The AM Forum

I've got a anan 10e coming this way to evaluate from a friend.

Always (the bit I've used sdr)  used flex and powersdr...  What software would you recommend...   For AM and ssb,  if different package by mode.

I'm contemplating a purchase,  and my friend bought one for a remote site.   Upon hearing I wanted one,  he offered his to try while  the remote is being built.

--Shane
KD6VXI

Shane, essentially the only game in town for any mode is the OpenHPSDR version of PowerSDR available here:

http://svn.tapr.org/repos_sdr_hpsdr/trunk/W5WC/PowerSDR_Installers/PowerSDR_mRX_PS_v3.3.6.0_Setup.msi (http://svn.tapr.org/repos_sdr_hpsdr/trunk/W5WC/PowerSDR_Installers/PowerSDR_mRX_PS_v3.3.6.0_Setup.msi)

73, Barry N1EU

Lol OK then.

He received the radio,  sent me a pic of it already.

Not so impressive,  just the box.   What it can do....   Wow.

Thanks!

--Shane
KD6VXI

Shane  I have a 10E, its a great rig for any mode including AM.  Many reviews out there so wont go into all that. But it does require a good performing Windows based PC and needs a modification for pure signal from a linear amp, nothing too difficult.  It is now my main rig, sold all my knob radios.  For AM it has a wonderful sync detector and the carrier ratio can be adjusted to give clean positive peak over modulation. Hard to beat for under 1K$.

73s  Nigel

I have an ANAN10E, which I use in conjunction with a 100W Elecraft KXPA100 solid state linear amplifier.

I have posted information on the Yahoo Groups Apache Labs web site regarding how I use Pure Signal without having to make any modifications to the ANAN10E.

When transmitting, the "RX" SMA connector on the back of the ANAN10E provides an input to the ANAN10E's receiver... even though the ANAN10E's schematic shows that there is a relay (k19) that places a short to ground across the input of the ANAN10E's receiver... on transmit. The reason is that, inside the ANAN10E, there is a short length of 50 ohm coaxial mini cable between the input of the ANAN10E's receiver and relay k19. The impedance... looking into the receiver end of this coaxial mini cable... is not 0 ohms when k19 is in the transmit position. It is actually (using a Smith chart to calculate the impedance of a shorted length of coaxial cable) about j5 ohms on 40 meters, j2.5 ohms on 75 meters, j10 ohms on 20 meters, etc.

I place an external 5W non-inductive, 50 ohm resistor, located adjacent to the ANAN10E's RX SMA connector,  in series with the RX SMA input connector.

The result is (for example, on 40 meters): the impedance... looking into the added resistor that is connected in series with the SMA RX port... is very close to 50 ohms (resistive). I.e. 50 ohms + j5 ohms is approximately 50 ohms. In addition, the combination of the added 50 ohms of series resistance and the impedance looking directly into RX the SMA port (i.e. j5 ohms) forms a 20dB voltage divider [i.e. j5/(50+j5)] feeding into the receiver of the ANAN10E on transmit.

In the case of my setup, I drive the input side of the added 50 ohms series resistor (located adjacent the ANAN10's RX SMA connector), using 50 ohm coaxial cable... with a 10:1 current transformer that is sampling the output of my 100W amplifier. The secondary of the current transformer has a fixed, 150 ohm, non-inductive resistor directly across it (for safety)... so the parallel combination of that 150 ohm resistive load, and the 50 ohm resistive load looking into the coaxial cable that feeds the RX SMA input of the ANAN10E, produces a total load across the secondary of the current transformer of 37.5 ohms. The voltage reduction between the output of the amplifier and the output of the current divider is, therefore, (1/10) x 37.5 ohms/50 ohms = .075 = (approximately) 22.5dB.

The net voltage attenuation between the output of the amplifier and the input of the ANAN's receiver is (approximately) 22.5dB + 20dB = 42.5dB. That works fine with my 100W amplifier, for providing a proper feedback signal for Pure Signal. I.e. with this level of external feedback, Pure Signal sets the ANAN's internal adjustable receiver input attenuator to 17dB. The total voltage attenuation between the 100V peak RF output voltage of the amplifier, and the input to the ANAN receiver's RF preamplifier (including the attenuation of the ANAN's internal adjustable receiver input attenuator) is 42.5dB + 17dB = ~60dB. Therefore, the Pure Signal input to the receiver's RF preamplifier is 100mV peak. The ANAN receiver's RF preamplifier has a gain of 20dB... so the Pure Signal input voltage to the A/D converter = 10 X 100mV peak = 1V peak.

If you have a 1000 watt output amplifier chain, then you need an additional 10dB of attenuation in the feedback path.

Stu  

I have one on the way as well.
I worked Stu (AB2EZ) on his and it works better and sounds better then his ranger.

I had a cheap computer I got at wall mart (AMD quad core) with a crappy video system and it ran the flex 5000 and 3000 fine, almost no latency.
I replaced it with an I5 computer with a much better video system so I expect it to run the 10E fine.
All I do with the computer is to run the sdr, it is not connected to anything else (internet) and its not doing anything else, and they have always been stable and well working that way.

I expect to just run one RX at a time on the 10E, should be easy on the computer.

I read the owners manual (on line) for the 10E and they say AM is 3 watts carrier max.
They say the duty cycle matters and that that the output transistors are good for much more power but the rest of the parts are crap....

3 watts is not a lot.

Is there any reason I can not use the mod monitor take off to feed the signal in for pure signal?
I also have an attenuator box (push buttons) made by HP to adjust any level if the signal level is too high.
My mod monitor takeoff is a 10KV 20 PF variable cap in the station control that comes off the coax.
Is there a level indicator for the 2nd receiver input?
What range is allowed?

Brett

With 100% positive modulation peaks, 3W of carrier power corresponds to 12W of peak power.

I set the "carrier level" to 75 ... which corresponds to a carrier output power level of:
0.75 x 0.75 x the nominal AM carrier level. The nominal AM carrier power level is 0.25 x the peak power output level that you have set.

Therefore, if the peak power output level is set to 12W, then the nominal carrier power output level is 3W.

Examples:

With the "carrier level" set to 100 (the default setting), the actual carrier output power level is 1.00 x 1.00 x 3W =3W = the nominal carrier output power level

With the "carrier level" set to 75,  the actual carrier output power level is 0.75 x 0.75 x 3W = 1.7W

With the "carrier level" set to 80, the actual carrier power output level is 0.8 x 0.8 x 3W = 1.92W

A "carrier level" setting of 75 allows for positive modulation peaks of: 100% x (2-0.75)/0.75 = 167%

A "carrier level" setting of 80 allows for positive modulation peaks of: 100% x (2-0.80)/0.80 = 150%

A "carrier level" setting of 100 allows for positive modulation peaks of: 100% x (2-1.00)/1.00 = 100%

Stu

Right, and it does not get hot, does it?

Brett

Whether you use the approach I described in my earlier post (adding an external 50 ohm non-inductive resistor in series with the ANAN's RX sma input port), or you make the modifications to the ANAN that are described on the Apache Labs Yahoo Group Web site:

The interface between the output of the amplifier and the ANAN's Pure Signal input port will be looking into a 50 ohm impedance... and it should be designed accordingly.

Your modulation monitor's  RF sensor/sniffer interface to your RF output signal may or may not be able to drive a large enough sample of your RF output signal into a 50 ohm load... to satisfy the requirements for Pure Signal to work properly.

With the approach I described in my earlier post, the RF sensor/sniffer must produce approximately 1 watt (peak power) into a 50 ohm load.

If you make the modifications to the ANAN that are described on the Apache Labs Yahoo Group web site, then your RF sensor/sniffer must produce approximately 10mW (peak power) into a 50 ohm load.

Stu

When I run my 10E on AM I use 2.5W carrier with a 75% ratio. This drives my HB amp to 8A at 50V or 400W input and carrier OP about 250W.  The 10E gets a little warm but certainly not hot.  The amp fans go to high speed after a bit but it does not overheat either. The signal reports are all good.

73s  Nigel

I'd be careful about lowering the carrier level setting unless you've got a modulation monitor on the output.  Current versions of PowerSDR do an excellent job of full AM modulation.

73, Barry N1EU

They have a web site?
I know there is a yahoo forum, but no sign of the web site?

I guess the correct answer Brett is that there's lots of information spread across various web sites. Barry - N1EU has collected a lot of links and information at the wiki site located at the link below. It's a good place to start, especially for the articles that cover setting up AM operation:

http://anan-100d.wikidot.com/ (http://anan-100d.wikidot.com/)

Regarding Shane's question for the best software. I would agree that the OpenHPSDR package really does have it all. After you have mastered that you might find it fun to mess around with the Zeus Radio software. It is not as full-featured as OpenHPSDR but it does an excellent job with TX and RX on AM and has some clever features that are implemented in the TX audio chain. The video below shows how they handle the user interface for adjusting gain distribution. Note that the audio heard in the video is what is generated by the Zeus Radio software:

https://www.youtube.com/watch?v=-0IFU2bcGX0 (https://www.youtube.com/watch?v=-0IFU2bcGX0)

There is no real installation involved as you simply unzip the software into a folder and run it from there and it does not interact at all with your OpenHPSDR installation. The latest version is at the link below. Just click on the downward pointing arrow at the top of the page:

https://drive.google.com/file/d/0Bylg7ZUFCyArS3I5dEhoOWlIMzA/view (https://drive.google.com/file/d/0Bylg7ZUFCyArS3I5dEhoOWlIMzA/view)

There is an excellent quick-start guide in the files section of the KC9XG user group at Yahoo. You can get that by joining the group or if you want I can email the startup guide to you (1 meg PDF file).

73,

Rob W1AEX

I am used to PowerSDR from the Flex stuff, so that is a start.
Got the OpenHPSDR software installed yesterday and went though the usual settings.
Now I just have to wait for the radio.

Out of all the software I have used, PowerSDR is up there, its easy to use and has the buttons you want handy.
The Flex newer versions were very fussy about audio levels.

Brett,

Although the GUI for OpenHPSDR looks similar to PowerSDR everything under the hood is new. I think you will enjoy the TX audio chain which has a look ahead algorithm that makes it impossible to drive the ALC beyond 0 dB. You can drive it hard (in fact you need to make sure you are driving the ALC to 0 dB regularly) and the TX audio will get loud and remain clean. The leveler also functions as a software limiter that allows you to bring up the audio density. It's a snap to set the audio gain distribution. You'll also find that the AM detector is distortion free. I'll be interested to see what you think after you've had a chance to mess around with it.

73,

Rob W1AEX

Rob,
I have been reading a bunch of stuff you posted about it, seems they fixed a lot of the issues the flex stuff had.
I know you could get the flex stuff to sound clean but not do tricks like the new stuff can.

I received the 10E today.
I am not impressed.
The Ethernet interface comes up at 100 mbps which is enough I suppose, supposed to support 1 gig.
The radio only puts out 2 or 3 watts carrier and 6 watts pep at the most.
Latency is no better (worse I think) then a Flex 5000 running with a firewire interface.
I can get the latency low with very small buffers but then thing go south performance wise.

Receive performance seems fine, I did not really get a chance to compare it to other things on weak signals.
Flex 5000 performance is very close I think, at least in most aspects.

Not overly fond of power sdr mrx, I like the control and options and better displays of sdr-radio version 1.5, it allows you to  select which sideband to listen to with the click of a button.

Audio (RX and TX) sounds very good, typical SDR there.

So $1000.00 for a 3 watt radio with a lot of latency seems like a poor investment, I still like the SDR-IQ better as a band scope and a receiver.
A 3 watt transmitter with a lot of delay in it I have no use for.

Brett,

Yes, even though the onboard ethernet hardware is a 1 gigabit port, the driver is currently limiting it to 100 mbps. This is holding the total number of available receivers to 5, but since the 10E only has 2 receivers it does not make a difference there. However, it does eliminate the use of stitched receivers, and the 384 kHz sampling rate when pre-distortion is enabled. The developers do have a 1 gigabit driver running but have not released it to the general public yet.

Regarding the low power output you are seeing, if you have not yet done the PA calibration for each band it should be possible to quickly bring that up to spec. The 10E owner's manual makes reference to this on page 38 but doesn't actually explain the procedure. All you do is observe your power output while going band-by-band and hitting the Tune button in OpenHPSDR with the Drive slider up to the maximum. If the power output is not to spec on any band, in the PA Calibration table adjust the gain setting for that band downward in value and this should bring your output power up.

The latency issue you are experiencing is a puzzle to me. My own is extremely low for TX and RX. I haven't bothered to measure it because it is not problematic, but if I listen to a station in the AM BCB with a standard radio, and then tune the same station in with my ANAN SDR, the latency is just barely discernible to the ear and is far less than my Flex 5K which ran through the firewire port. The first question I would ask is what is the reported CPU Utilization in OpenHPSDR for your system? My own runs between 4% and 8% typically.

You should be seeing a far cleaner and quieter receiver without the LO and solid-state relay noise issues that are present in the Flex 5K/3K/1500 generation of radios.

Simon Brown is currently working on his SDR-Radio software for the ANAN. He is currently using it on 60m (TX and RX) with his own ANAN 10E but is not making any promises yet about when it will be ready for primetime.

Definitely hit the apache labs yahoo group for support for anything you feel is unsatisfactory.

73,

Rob W1AEX
  

Rob,
Thanks for the response.
My cpu runs around 10% (I5, 8gb ram windoze 7)

The latency I am talking about is on TX.
And, if I transmit on another (analog) rig and listen in the 10E there is delay.
Its ok on the smallest RX buffer size.

The radio does 10 watts out on CW and I would expect it to do 10 watts pep, but it looks to be about 6.

I was doing some a/b tests with receivers, I have a homebrew bridged into the antenna and both radios go into the marantz audio amp, and the homebrew seems to recover audio a bit better on weak signals but its close.


I expected less latency through the system over ethernet.
It does have much less then the sdr-iq on usb but a good size buffer selection kills it.

Brett

On the subject of latency... i.e. the time delay between the RF input of the receiver and the demodulated audio output:

In any receiver... whether it is a traditional receiver that incorporates analog filters, or a modern receiver that incorporates digital filters... the time delay between the signal going into the filter and the signal that comes out of the filter must be at least 1/[2 x pi x df]... where "df" is the change in frequency that is needed to transition between the end of the passband (e.g. less than 1dB attenuation) and the beginning of the stop band (e.g. greater than 6dB attenuation).

If the filter "skirt" is so steep that it rolls off (passband=> stopband) when the frequency changes by only 5Hz, then the delay through to filter must be at least 1/[2 x 3.14 x 5Hz] = 32 milliseconds.

This is a mathematically provable lower bound on the delay through any filter.

In the case of an SDR, if you set the bandpass filter to have steeper rolloff, then the latency will increase proportionately. If you set the bandpass filter to have a more gradual rolloff, then the latency will decrease proportionately.

In a classical receiver, the LC or crystal or mechanical i.f. filters would typically have a moderately steep rolloff. For example, if "df" is around 100Hz, the delay through a classical crystal or mechanical filter is around 1/[2 x3.14 x 100Hz] = 1.6 milliseconds.

If you want low latency, don't make the digital filter rolloff too steep!

[Another way to think about this is as follows. The steeper the rolloff of a digital filter, the larger the number of most recent samples of the filter's input signal must be collected (buffered) in order to perform the required digital signal processing. If the digital filter's sample rate is 48kHz, and you want the digital filter to roll off between the passband and the stopband in "df"= 12Hz, then you need to collect, buffer, and process the most recent 48kHz/(2 x 12Hz) = 2000 samples of the input signal. With a sampling rate of 48kHz, it takes 2000/48kHz = 42 milliseconds worth of recent samples to perform the digital signal processing. This delay is 3.14 (i.e. pi) times as long as the minimum delay described above]

Separately, any digital audio device that includes a buffer will add a delay that is: the size of the buffer / the sampling rate.

For example, if the buffer size is 1024 samples, and the sampling rate is 48000 samples per second, then the delay through the device will be 1024 samples/48000 samples per second = 20 milliseconds.

None of the above has anything to do with the plumbing (e.g. usb v. Ethernet), or the processing speed of the computer/FPGA/microprocessors used to perform the digital processing. It is possible for the interface between the computer and the SDR (e.g. usb or Ethernet) to add delay... but that delay is in addition to, and usually much smaller than the delays described above.

Stu

Latency is the nature of digital signal processing and there's no way around it currently.  Both the ANAN and Flex 6K (and Flex classic) are burdened with it.  Flex 6K works around the latency by providing a tx audio monitor that is taken off before most of the software processing.  Depending on mode and filtering, we're probably talking about 50-150msec of latency on ssb rx/tx and cw rx.  Both ANAN and Flex have lower latency for cw tx.

As Brett suggests, the ethernet transport contribution to latency is a small fraction - it's the software.

73, Barry N1EU

Unless the driver is complete crap, the latency on a local (no switches involved) Ethernet, as in the ANAN, should be in the microsecond range.

Let me know what you want for it as I'm looking for a second one.

Nigel

Right, the latency over the ethernet connection is so slight it is not a factor for our purposes. My TX latency on SSB/AM is ~40ms which is not enough to bother me if I monitor my own audio.

Beginning with OpenHPSDR version 3.2.24 the developers started implementing a new "software/hardware communication protocol" that unhooked the internal DSP sampling rate from the user selectable sampling rates in the main audio tab. The internal DSP is now locked at 48k during RX and 96k during transmit. This allows you to run much lower buffer sizes in the user selectable fields without affecting the performance of the internal DSP functions. There is no longer a need to change your DSP buffer sizes to maintain the same filter shapes when you change your sampling rate in the primary audio tab. Warren Pratt (NR0V) now suggests using a DSP Phone buffer size of 1024 for RX and no more than 2048 for TX as a starting point. Barry - N1EU documented some of Warren's suggestions at the ANAN-100D wiki site:

http://anan-100d.wikidot.com/100d-buffer (http://anan-100d.wikidot.com/100d-buffer)

At any rate, this allowed many users to experiment with sampling rate and buffer settings to attain much lower TX latency than before. I had no issues running a primary audio buffer size of 512 and a sampling rate of 192kHz for phone and an RX buffer size of 512 and TX buffer size of 1024 in the DSP Phone buffer settings. No drop outs or anomalies of any kind unless I ventured lower. Eventually, I settled on a primary audio buffer size of 1024 and DSP Phone buffer settings of 1024 for RX and 2048 for TX because I had a feeling that the leaner settings were pretty close to the edge and the increase in latency was not enough to bother me. Scott - WU2O has done a lot of experimentation and has had success with very aggressive settings (such as 256 for the Primary audio buffer) when he played around with getting his system as near to real-time as the software would reliably allow when he worked with external audio processing DAW/ASIO software.

One thing that most AM users have noticed is that in the DSP>AM/SAM tab it is a good idea to uncheck the Fade Leveler box as it can cause a long delay in recovery time when going from TX to RX. It can also cause issues when you are receiving a very strong AM signal.

Your I-5 with 8GB of memory running Win 7 should have absolutely no issues with handling OpenHPSDR. The CPU utilization you are seeing seems to be right on the mark so it certainly doesn't look like a computing horsepower issue.

Rob W1AEX

    Brett,

   I wonder how you are measuring power? Most power meters are rated at something like +/-5% or even as high as +/- 7.5%, and that is at full scale! A watt meter measuring 10 watts on a 200 watt FS range can be easily +/- 50% off and not be out of spec since there is no accuracy guarantee anywhere except full scale. Then there is that PEP stuff where every meter designer more or less integrates that peak, and gives you a number. I'd consider it a coincidence if two PEP watt meters read exactly alike with speech. The MFJ un-powered meter I have shows PEP at 2X carrier when the oscilloscope shows 100% upward at 2X the amplitude of the unmodulated carrier (4X the power).

Jim
Wd5JKO

First off, I adjusted the power calibration (for 40 meters only) and can get 5 watts carrier and 20 watts pep out, so its possible to set things however you want till the radio blows up (or melts down).
4 watts carrier and about 20 watts pep is ok with the AM carrier set to 80%.

Sounds good once I put the TX eq on.
I put all the buffers as small as they will go and there is still enough delay to make me sound like I have brain damage.
I tried the built in monitor and the mod monitor and can get the delay almost low enough to be ok.

Made a few contacts barefoot on 40 meters (5 watts out), one with WB9AGK, Indiana is a good hop from NJ.

It seems like a good radio to put into a solid state amp and run at 50 to 100  watts carrier out if you can deal with the delay (not monitor yourself).

I expect Ethernet to be quick and low latency, maybe some computers process Ethernet differently so I will test the radio on other computers.
USB ALWAYS seems to have a good amount of latency, after all, you plug all sorts of things into the bus, mice, keyboards, radios. There is a big buffer in USB I expect.

Firewire was quite fast and low latency, maybe bandwidth limited.

In all cases, the transmit audio sounded clean, and the receive is the typical SDR high fidelity.




 

I agree with Jim

I once talked with a fellow whose small company produced a digital wattmeter which displayed "peak power". I asked him why he didn't design his meter to actually capture and display the true peak envelope power of an AM or SSB signal when the modulation bandwidth was around 3kHz... rather than displaying a much lower peak power number (about 60-70% of the true PEP).

He told me that customers were used to buying peak power meters (like Bird model 43 with a peak power adapter board installed) that only read about 60-70% of the true PEP when driven by a voice modulated SSB or AM signal. He said that his customers didn't want a meter that reads the true PEP because his customers want their true PEP to be higher than 1500W... and they want a meter that (falsely) shows that they comply with the FCC PEP rule.

Stu

    Perhaps you can tap off your audio going into the SDR and use that to drive a headphone amplifier. The latency will be essentially zero, and you will hear yourself just fine.  :P

I find it interesting how much delay there is in over the air digital TV. Satellite is way worse! It is what it is.. Cannot use local AM sports station to get the audio to replace the TV audio...too much delay. :-(

Jim
Wd5JKO

I can not say exactly how far off the peak reading is on my watt meter, I can only compare it to other transmitters.
Most will do 4 times carrier, and the meter itself has three scales, 20, 200 and 2000.

I noticed an odd thing on my radio, I can adjust the radio to do 5 watts carrier and 20 watts pep indicated on my watt meter on 40 meters, but on 80 meters I get 10 watts carrier and 15 watts pep using the same settings.
I have to turn the AM carrier level down to about 10% to get 3 watts carrier and 15 watts pep, while 40 does 5-20 watts at 75% carrier.
Maybe some sort of band pass filter tolerance...

Stu

We've had this discussion somewhere in the past and it squares with my experience in my station.  Here are three slides that I use to demonstrate the relationship with peak reading on a scope, my Bird set to average and finally it set to the peak reading setting.  There is one slide missing: the one where I set my carrier to full scale on the scope while watching my Bird for a 1500 watt carrier level. All this while using my 1500 watt load. 

Al

Brett, I heard you on the air over the weekend. Sorry I was not able to jump in. While you had enough audio, it was no where near your normal sounding audio - definitely lacking highs. I had my 6500 opened up to 10 khz. Just curious why a brand new radio requires you to do a transmitter alignment? A friend of mine had one of the higher model Anan's and he found that the IMD spec was unacceptable until he did this alignment. Poor QC? It's kinda scary (to me) when a you risk the possibility of damaging the radio if the carrier is set too high. In the end he decided to go with a Flex 6700 - far less tweaking, more time to enjoy the radio!

Re-read Rob's no. 24 and preceding posts (good explanations) after Anan's entry into building the Hermes card and with others' subsequent software enhancements.

-- then :
It's a calibration procedure to marry Flexs' legacy software via the now modified Power SDR
for use in driving a garden variety 20 watt pep amplifier at 12 to 15 watts or whatever you choose.  Since the gain varies for each ham band, you set each band at say "39db," or "X" etc. to achieve a constant output of your choice. Most guys set it at no more than 15 watts.

Naturally at 10 watts (The nominal sales rating), the IM, etc.
will be considerably less than at 20 watts.

Here's a nice description of the Hermes board written in pre Anan times by N9VV.
http://n9vv.com/Images/Hermes/Hermes%20an%20Introduction%20October%202011.pdf

Also I sure liked Stu's and subsequent filter latency explanations.  Stuff to remember about "free lunches. "

Part of the time running the radio on the weekend the TX eq was not turned on.
I may have forgot to click the box or it reset itself...

I can understand the needing to set the gain of each band, flex would likely do that for you while the ANAN stuff is not really being sold as a finished rig.

But why the big difference in the way the radio works between 40 and 80 meters?

Its not just output, but peak power and the ratio of peak to carrier with the same settings.
40 meters seems to act just like it should, and 80 meters is all wacked out (but still sounds ok).

Latency is expected, but at what level?
500 ms is getting bad, I can get it to a full second with buffer changes, or maybe down to 250 ms with the smallest buffers (almost ok).
The flex 5000 would do about 100 to 150 ms without the buffers being the smallest.

I will test the radio on other computers...

Brett,
You have probably already done this but have a look at your setup parameters and compare them band to band. Once set and saved on 75 meters you may find settings are not the same on 40. This includes main screen drive settings.

Robert, SQQ
Doing a software calibration has nothing to do with a transmitter alignment or product quality control. The Flex Radio equipment, from the SDR-1000 to the 5000 as well as ANAN products all required calibration. Using a new rig without calibration is not recommended.

Mike

Brett, Perhaps the early Flex software written with softer filter attack and release times as per Stu referenced calcs.

Was your experience real tr/Rd delay in QSO with Someone?
Were the earlier Flexes picking off audio sampling before the heavy digital lifting as already suggested and as in some instances still occuring in Flexes?

You had this problem with QS1R too as I recall.
Interesting.

Brett

At least two things are definitely wrong with your observed ANAN10E behaviors:

1. I have a 10E, as you know, and I have carefully measured the latency by using a digital oscilloscope and also a Symetrix 322 digital signal processor, as a delay line, to compare (and also to align in time) the 10E's audio input to the 10E's demodulated RF output.

The measured delay between the audio input and the demodulated RF output (observed with my REA modulation monitor) is approximately 68 milliseconds. Note that the audio is being applied to the microphone input on the front of the 10E, and this measurement is, therefore, being made without the use of VAC.

I'm using a 1024 sample transmit buffer (see the setup => dsp => options window) on phone. Note that 1024 samples / the internal digital filtering sample rate of 48k samples per second = 21.3 milliseconds of delay (latency) associated with the transmit digital bandpass filter.

If I change the transmit buffer length from 1024 samples to 512 samples, which reduces the steepness of the bandpass filter skirts from around 47Hz (passband to stopband) to around 94Hz (passband to stopband), the total delay (latency) decreases from 68 milliseconds to approximately 54 milliseconds.

Note: The DSP settings (e.g. the transmit buffer length) are set (and saved) separately for each "transmit profile". So, for example, if you change from 3100Hz SSB to 5000Hz AM... make sure that the DSP settings are what you want them to be.

2. My 10E has a peak output power on all HF bands that I have consistently set to 12W ... using the "tune" mode, with the drive set to 100 on each band, and the transmit power calibration settings individually adjusted to between 50 and 39 on each band. It appears that you have something set wrong ... that is causing your power output to be low on 80m.

The improper setting could be any of a number of things... including the 80m drive setting or the low pass filter settings (setup => general => ant/filters => LPF).

The drive level is set (separately for each transmit profile) by a combination of the "drive" control on the main GUI display, and the drive-related settings in the setup => transmit window. Make sure that the "tune power" box is set to the tune power level you wish to use, and that the "use drive power" box is not checked. Alternatively, check the "use drive power" box... and then use the main GUI display's "drive" control to adjust the drive power. 

The LPF settings are supposed to be set automatically via the firmware... but it is conceivable that the 80m setting is not correct.

Separately, in the setup => general => hardware configuration tab... check what the ANAN10E's firmware version is (lower right corner of this window). If the firmware version is not 1.3, you might want to go through the process of flashing an updated (i.e. version 1.3) firmware version into the ANAN10E's RAM/FPGA.

Stu

Brett,
Also make sure you haven't inadvertently set a large xmit latency as per:

Mike I've owned the Flex 5000 and now the 6500. Aside from checking the frequency calibration (with WWV) I've never had to do any alignment or calibration. I've created profiles, but these were done for operating convenience. I think the Anan radios are incredible, but they seem to be more about constantly adjusting, updating and experimenting rather than operating the radio for the enjoyment of the hobby. No argument, they do sound very nice!

I think you are mixing up the delay between going from RX to TX to audio delay through the system.

My problem is when I speak into the microphone, the audio in the mod monitor or the radio built in monitor is delayed enough to be unusable.

The screen you are showing is the delay time between switching from RX to TX.

And I must say, I did not expect the radio to be plug and play, all sdr's take some setting up as you have control over so much, every aspect of the radio can be adjusted which is a good and a bad thing.

I never had to do any calibration with the Flex stuff, maybe it would have been a good idea but they seemed to get close enough out of the box to not need it.

Part of the fun of this stuff is you get to play with things, lots of things, without letting the smoke out (most times).

And the QS1R was a USB based receiver and so has all the USB latency built in.
I have run an sdr-iq on many different computers, along with a flex 1500 and other USB based radios, and every radio had a lot of delay on every computer if it used USB.
The only radios (so far) I got the latency low enough to use with a monitor was the flex 3000 and 5000 on firewire, and not even on really fast computers.

There MAY be big buffers on some Ethernet ports/cards and not others...

Oh, ok.
Well fwiw I hear my voice on PowerSDR monitor through phones right from the Anan 10 just slightly behind from what I speak.  I'm running an old ASRock Atom computer. Delay Just enough to be a slight echo, not enough to get tongue tied. Sounds like someone with just a bit of echo from their audio lashups.

On some settings of the Qs1R it was enough to actually hear a couple of words for what was true RF modulation going to the dummy load.  ;D
Actually good for what I sounded like, just like playing back a tape.

All the Power SDR default settings seemed to do ok for first plug-n-play trials. 

Well keep plugging away. You'll find it.

Personally, I prefer to have all the calibration tools available to me. This is one of the things that irked me about my Flex 5000. After the v2.x.x versions of PowerSDR the PA calibration routine was locked down in the hidden routines and if your power output was not to spec you had to send the rig back to Flex for calibration. The PA calibration was only possible with an embedded software routine in a hidden menu and required a code file from Flex to be placed in a particular folder and an LP100 meter connected with a dongle to run it. Fortunately, I kept my old versions of firmware and vintage PSDR versions so I could follow the convoluted instructions to go backward with firmware and successfully calibrate mine, but most guys were stuck with sending the box back to Flex. It's so simple to run through the PA Calibration table in OpenHPSDR and set each band for 100 watts output using the tune button with the drive slider set to the max. It takes less than 5 minutes and once you've done it that's it, no big deal. Take a screenshot of the PA table and keep it for future reference when a new version of OpenHPSDR comes out.

I have never heard of anyone having the issue you described when comparing AM output between 40m to 80m. The ratio is locked to the transmit profile, not the band. As someone said, you might not have the PA Calibration setting adjusted correctly for the band that will not produce the expected 4:1 ratio between PEP and carrier. I have found an AM Carrier Level of around 80 to guarantee me around 120% positive peaks across all bands. One thing to note is that if you enable pre-distortion while running AM the asymmetry will go away. I think the algorithm interprets it as distortion and so it corrects it.

I'm not sure why you are seeing such horrible latency Brett. I'm using a UMC202HD interface to run my hardware audio chain into the computer through a USB port and use VAC to run that into OpenHPSDR and I get results similar to what Stu measured. I can easily tweak it even tighter to get it down another 10ms or so, but it really isn't even discernible the way it is currently set when monitoring my own TX audio. Maybe you could post some screenshots of your Audio>Primary tab and your DSP>Options tab for one of your AM profiles. It might be something simple.

73,

Rob W1AEX

I hate to throw water on the new state-of-the-art SDR's and their exotic design and wonders. Not singling out any particular manufacture. But spending all of that money on something that is flea power like 3-15 Watts of power is major RIP.
And then folks looking to pay another frickin $1500 for an amp to bring it up to 100 watt final. Crazy.
Now I see a thread where someone wants to amplify the flea
power from an Anan 10 to something usable. Geeesh

Just my .002 cents worth of criticism.
Fred................feeling a little feisty today in the hospital

Mopster,
Glad your feeling a lil' feisty today.  Great! and keep em coming.
We're rootin' fer ya.

Bret,
A pix to warm Fred's heart.
Here's a screen shot of Latencymon on my old beat up Atom330 with PowerSDR running.  The ol' computer hain't so hot it seems but works ok with reduced, say up to 96k 'bandwidth.'

If you haven't already this or similar program, get latencymon v 6.5 home edition (fware) from
http://www.resplendence.com/latencymon

Edit : and another screen without PowerSDR or internet running, just all the windows junque and AVG.

Well, the sdr's can get really good audio out, like if you send a square wave into the box, that is what comes out the antenna, even at 20 Hz.
No, you do not need that, but the ability to do it is interesting.
Some people like the idea of perfect.

And don't forget, you get a world class receiver built in.
There is no analog receiver that comes close.
Bandwidth, filters, distortion levels, the band scope, you can spend $20,000.00 on an analog receiver and not get as good of a performance.

I do not have any use for a 5 watt or a 25 watt radio other then fooling around with it, it would never be my main go to radio, but plenty of people use all sorts of radios into an amplifier, even silly ones like a dx60.

You could use one barefoot though, rough on the person at the far end maybe, but Stu has fun even with a poor indoor antenna at 20 watts carrier.
I worked Bob, WA2SQQ on 80 meters tonight with 5 watts...

I hooked up my old computer (quad core AMD) that used to run the Flex 3000 and 5000 ok and downloaded the software for the ANAN.
Ran at around 20% cpu, same latency through the system.
The cpu usage is higher on the ANAN then it was on the old Flex stuff.

The difference between 80 and 40 meters was the same.

I went through the process of setting the tune power output to 15 watts on each band from 160 to 20 meters which was easy and quick to do.
The difference between 80 and 40 meters was the same, high carrier power output on 80 and low peak power.
Same 15 watts out in tune, ssb could get to 15 watts pep.

I tried sample rates of 48k to 386K, all the different buffer sizes, and with the smallest buffers it gets to reverb levels.
Of course its fine if you do not monitor yourself, some people do, some do not, and I suppose it would be fine not to since you can see what goes out on the screen.

I worked Bob and he said the audio sounded very bassy, so I cut it all the way and boosted all the highs and it sounded better. Old but nice radio shack 600 ohm balanced dynamic microphone.

Has anyone got the quick record and playback to work well?
Used to work great on the Flex stuff, seems to be mostly noise on the ANAN.

The firmware is 1.2 in the radio, not sure what the firmware does.

Every 100W transceiver includes multiple stages of amplification. I.e: a buffer amplifier (around 1W output), an intermediate power amplifier (about 10W output), and a final amplifier (100W output).

There are also some newly introduced, commercially built, ham radio amplifier products that incorporate new types of FET devices that can produce close to a kW out with only a few watts of input power. (They are currently not FCC compliant unless they ship with an installed input attenuator that the customer can remove).

Apache Labs makes a (nominally) 10W output transceiver (the ANAN 10E) with a $995.00 list price. They make a (nominally) 100W transceiver (the ANAN 100B) with a $1995.00 list price.

They are essentially identical, except for the inclusion of a 100W solid state final amplifier on the 100B's "amplifier/filter" board.

I purchased an ANAN10E directly from ANAN in January for $895.00 (the introductory, new product, "pre-order" price), and I purchased a separate Elecraft KPXA100, 100W solid state amplifier, directly from Elecraft for around $700.00.

The Elecraft amplifier uses the same pair of output transistors as the ANAN100B uses.

I prefer a separate transceiver and amplifier. There are fewer thermal management problems, the amplifier will not lose resale value over time as quickly as the transceiver will, and I can use the 100W amplifier with other QRP transmitters/transceivers (like my FT-817 portable transceiver).

Stu

Brett

In the ANAN products, the FPGA does all of the high speed digital signal processing, and also performs many of the on-board device control functions (like controlling the relays that select the low pass filters, push to talk, T/R switching, ...). The firmware is the set of software instructions that is used to program the interconnections of the gates in the gate array... and, therefore, the processing that is done by the FPGA. The firmware is stored in a RAM on the ANAN10E's "Hermes" board, and is loaded into the FPGA when the ANAN10E is powered on.

Audio processing, and processing to produce the displays (e.g. panadapter display) is done by the PC.

Again, there are a number of settings that are set and stored separately for each band. Something appears to be set wrong when you select 80m as your operating band.

Stu

What firmware is current?
My radio has 1.2.

Brett,

It looks like the newest is 1.3 for the 10E.

Rob - W1AEX

https://apache-labs.com/instant-downloads.html (https://apache-labs.com/instant-downloads.html)

Another thing I do not get is the radio seems to have very little or no latency on CW.
In the old flex stuff the sidetone was generated by the radio and reflected the delay through the system, and it was just like voice, very hard to send when the sidetone was .5 or more second behind the key.
The ANAN seems to have no real delay and seems to work like a normal rig on CW.

Brett

With my ANAN-10, I measure:

1.  Delay between pressing the paddle, and the monitor tone produced at the front panel headphone jack: not perceptible (no problem sending CW at 25 WPM)

2. Delay between the monitor tone produced at the front panel headphone jack and the RF output (as produced by my off-air monitor): 16 milliseconds.

This is controlled by the "key down" CW delay setting in the General => Options setup window. I have this set to 16 milliseconds

If I set the "key down" CW delay to 0 milliseconds, then there is 0 milliseconds of measured delay between monitor tone produced by the front panel  headphone jack and the RF output (as produced by my off-air monitor).

In CW mode (regardless of the setting of the DSP CW "transmit buffer" size), it appears that there is no narrow bandpass digital filtering being done on the transmitted signal. Essentially, when the key is closed, the high speed FPGA produces the digital representation of a pure tone at the desired RF frequency... and that feeds directly into the D/A converter and the RF amplifier chain.

In SSB or AM mode, there is delay (as previously discussed in this thread) between the audio input and the ANAN10E's "monitor" output (or, an external off-air monitor)... because (among other things) the monitoring is done after the steep skirt bandpass digital filtering has been applied to the digital representation of the output signal. That delay depends upon the number of samples being buffered in the bandpass digital filter (the DSP "transmit buffer" length), and on the delays of other buffers that exist in the transmit chain. With the DSP bandpass digital filter buffer set to 512 samples (in my case), that accounts for 512 samples / 48000 samples per second = 10.7 milliseconds of delay. Since the total delay (in my case) is around 50-55 milliseconds, the other buffers are adding about 40-45 milliseconds of delay. I'm not yet clear on what those other buffers are.

Stu

Feedback received this afternoon from Warren, NR0V, via the Apache Labs Yahoo group:

Hi Stu,

 Some months ago we moved the transmit CW generation from software to firmware.  This was done for just the reason you observed, i.e., to eliminate latency.  The CW Buffer setting for TX is no longer relevant for you.  No narrowband filtering is needed as (1) the waveform can be precisely formed with nice raised-cosine edges, and (2) everything is digital and the calculations are very precise.

 73,
 Warren  NR0V

Thanks for the info Stu.
CW used to be a problem with the Flex stuff, but the 10E works very well.

If you monitor yourself in headphones (built in monitor) it is not too delayed to speak normally?

Brett

As I have been told by one of the creators of HPSDR and the ANAN transceiver design:

On phone, the output of the ANAN's monitor is not a true representation of the modulation of the outgoing transmitted AM signal. This is an idiosyncrasy of the current design... and has (among other things) to do (if I understand it correctly) with the fact that the PureSignal pre-distortion correction is added before the monitoring point... but the monitor audio signal is obtained from the digital representation of the transmitted signal (before the D/A converter). It is not obtained from the receiver of the ANAN (which would add more latency).  So, what you hear with the monitor is the predistorted modulating signal... which is not subsequently compensated by the distortion of the RF amplifier.

I must admit that I don't know all of the details of this, but if I play a prerecorded audio clip into the ANAN's audio input (my own voice, calling CQ40) ... what I hear sounds different on the ANAN's monitor output v. my own off-air (REA) monitor. My off-air monitor sounds better.

With the latency now being only 50 milliseconds, I can monitor my AM signal with my REA off-air monitor while I am talking... and (after getting used to this slight delay) it is marginally acceptable. However, for listening to my own signal to check the quality of the transmission, I prefer to use a prerecorded audio clip as input to the ANAN. I usually don't monitor my transmitted signal in real time any more.

Stu

Thanks again Stu.
Then it sounds like everything is working normal here, I can get it to where I can monitor, the delay is just fast enough on the two smallest buffers that I can almost speak normally, that is its so quick its not impossible.

Any idea what happened with the firmware change from 1.2 to 1.3?

Its in the for sale section here.

Brett,
VK6PH the author is referenced in the Anan download if you didn't already know.

One can use the hpsdr svn to actually compare code changes line by line once gaining familiarity with the process.

http://svn.tapr.org/repos_sdr_hpsdr/trunk/Hermes/Release/

Hence:
http://svn.tapr.org/repos_sdr_hpsdr/trunk/Hermes/Release/Notes/Change.txt

And here's all the contributors. Fascinating wandering through the list. I've hit DL3HVH a bunch for CuSDR info and download.
http://svn.tapr.org/repos_sdr_hpsdr/trunk/

Glad you got your latency identified.
Same as Stu, I think I mentioned earlier that there was a very slight delay on SSB monitoring via phones from front panel output when using the Anan monitor compared with internal head speaking voice. --Just large enough to sound like a small amount of echo as if it were intentional.

Stu, than you tons for clear and concise explanations. It's been a lot of fun learning the ins and outs of the Hermes derived rigs.  I originally bought the Hermes board simply betting on CuSdr being Xmit completed a couple of months ago and then a Hammond case, but then decided to quickly exchange for the A 10. Supplier was game and quickly credited Hermes cost to the 10.

Thanks Rick.
I am not enough of a geek to get into the software that much, although it must be great fun to play with if you know how.

I had high hopes for the radio thinking the Ethernet interface and some of the work being done in the FPGA would reduce the latency to almost nothing, but it has not.
Perhaps on a really fast computer it could.

As it is, the ANAN is just a bit faster then the sdr-iq with its USB delay, and not quite as fast (or the same) as the early Flex stuff.

The QS1R was the same as the sdr-iq (in latency), along with all the other USB based radios.
Maybe one of the new RF Space receivers will be low latency, they all use Ethernet.

I will not be happy till its real time, and for a radio, I want 100 watts and a built in tuna.
Like all solid state radios, the 10E is picky about SWR and will not drive my 4x150's as it is, and will not even work into my antenna's if its raining.
Self protects above about 1.8 to 1.