The 6M-160M HPSDR Transceiver Project Begins

[K1JJ]

I played around with the buffers and yes, it does help the spike somewhat. But if I go to 256K I get drop outs.  I'll keep the DSP buffers about 500 and live with that.

I drove the FT-102 internal linear with the HPSDR. It uses three 6146's. I tapped into the pre-driver. It worked pretty well and was very clean. I estimate it was about -45db 3rd as the manual says. The side crud was about S5  compared to the Erb amp at S9+5.  The HPSDR alone is like S2.  

I was testing the Erb for about an hour and was working FB.  Then suddenly it stopped working. Looks like some gates opened up.   After seeing how good the tubes linear is, I'm going to go back to the TMC and continue working on that for now.  The Erb is too fragile for me to continue dicking around with it.  If you gets yours working super clean later on, I'll try again.

Update: Got the TMC working with a pair of class A1 6146's. VERY clean, even better than the FT-102 linear. Running AB1 requires the tubes to have no grid current, of course, thus the power out is low. I may have to replace the 6146's with a single grid driven class-A 813 to stay with those good IMD numbers.   Trying to get 30w out of a pair of 6146's starts to make garbage. 15-20 watts out pep seems the limit for -45db 3rd..  They will do 150W with heavy grid current, but IMD goes to hell.

T
T

[WA1GFZ]

Tom,
Try raising your screen voltage a bit to see if you can increase the power without grid current. Also might get away with raising plate voltage. Nothing like high Q tuned circuits.

[Steve - K4HX]

If your numbers are correct, the FT-102 setup is 40 dB cleaner than the Erb setup. How much more do you need? 

[WA1GFZ]

I've measured erb in the 30s but not all the way into class A.

[W2VW]

Tom,

     I'm interested in this because I have a mostly hacked TS-830 with the pre driver, driver and final intact. Was going to use it for the N.E.P.W. net.

R.U. just using a decoupling cap and load on the interface or what?

I'm currently getting 5 watts carrier from the HP8640B/30L-1/30S-1 ! Going for the record. Most power wasted to make a QRP signal.

[K1JJ]

If your numbers are correct, the FT-102 setup is 40 dB cleaner than the Erb setup. How much more do you need?  

You know how non-linear those FT-1000D S-meters are in the lower ranges.. :-)  I would say the FT-102 "linear"  is maybe 18db cleaner than the Erb SS amp in real numbers. That's quite a bit, actually.

Over time, I've compared many amplifier’s relative IMD using a normalized S9 +40 over on the FT-1000D or FT-102.  When using the 2.8khz position on LSB, tune up the band 3.5khz and while talking into the mike with nasty yallos, take a reading. If it is S9, then it’s borderline and on the verge of bad reports. If it falls into the S7 range or lower, this is about as good as BIG AB2 amplifiers will be.  The SDR transmit IMD is way down, actually stellar, like hovering at S1 to S2, if that much.  The FT-102 “linear” and TMC NFB amp using modified 6146's falls into the S5-S7 range.   Another test is to switch the receiver to USB on freq. The garbage should be very muted and hard to understand. Bad IMD will show S9+ crud and easy to understand. (this also depends on the receiver's opp SB supp, too)

I ran some IMD two tone tests using the Mercury spec analyzer and was able to put rough numbers to this.

Bottom line is I've learned a lot playing with this stuff and find it to be difficult to create a complete chain from exciter to QRO and still maintain ~ -45db or better.  Once you HEAR the difference between -30 and -45db on the receiver, you neva go back!  

Dave - I found the transistor just before the 12BY7 and tapped in with a series 150 ohms, then 50 ohms to ground and then .005 cap to the base. This required about 10mW for full output. You can go back one more stage if needed. Watch for the ALC action, since it might still be in the circuit, most likely.

I may take out the 6146's and put in a single 813 with 900 V plate, 300V reg screen and bias to make it class A. It will have heavy negative feedback.  All this just for ~40 watts CLEAN to drive the Dr. Love chain. Dr love chain already is -45db-ish, so had to get rid of the FT-1000D's -30db garbage.  Though, the FT-1000D seems maybe -40db 3rd when running at about 30w pep out.  So you will probably be OK, Steve, when you drive the new amp chain as we discussed.

Frank, you are right about the higher screen. I need to add another series reg to ramp it up for the 813. Since it must draw screen current, it is regulated.  I'm already at full plate  diss with the 6146's and see only 15w clean in class A1, thus the 813 is needed. [sigh]

T

[WA1GFZ]

Tom, The FT1000D at 40 over nine is generating internal IMD or your calibration is bogus

[WA1GFZ]

Tom how about a pair of 4D32s

[K1JJ]

Tom, The FT1000D at 40 over nine is generating internal IMD or your calibration is bogus

Relativity is what counts.  Even with the big amp on (Dr. Love) I see only S7 crud up 3.5khz.  Why would the 1000D generate internal IMD with a short piece of wire off the back?   You need at least  40db over signal in order to hear the low level IMD.  Bottom line is many of my tube amps in the past had IMD problems as reported on the air and heard it also in my test receiver. They both confirmed. Now that I have verified Dr. Love to be clean in the receiver, I have gotten no bad reports. I also verified it with Steve's high-tech spec analyzer last year when he dropped by at about -38 to 40db 3rd.   I have improved the rig for better figures since. In addition, the Softrock using a computer 2-tone test showed the same thing.  My receiver measurements do show the difference between a good amp and a problem.

Bottom line is the CCI amp and the Erb amps SHOULD be acting cleaner, but they are not for whatever reason, I dunno why.  The data sheet specs say they shud be -50db 3rd using my parameters. You can try both lashups over there if ya want, but I doubt you have the time right now..


BTW, I decided to put the 4CX-150's back into the TMC and make it stock again. I will run reduced voltages and find the sweet spots for 40w pep output.  It will make a good "barefoot" amp as well as a good HPSDR IPA for the Dr. Love chain.  Can't fight a good designm.  Gonna try a slow pressurized muffin fan under the tubes for quieter operation at low pwr.

T

[WA1GFZ]

Tom,
40 over S9 is about 95dB above the noise floor with a calibrated S meter.
Close in there is a good chance the receiver is generating IMD. Look at the Sherwood specs and you will se you are well above the point that would produce IMD. I have been tricked many times falling into this trap. The HPSDR Mercury is my best spectrum analyzer for dynamic range.

[K1JJ]

If there was bad receiver IMD, then it would show up equally on the Erb or the tube amplifier since both were normalized the same into the RX. Why would one look poor while the other looks good if it were not the case?

Besides, I have used the Mercury (and the Softrock) spec analyzers and they show basicsally the same thing as the FT-1000D. So I simplfy it and use the RX for now.  But later I will go back and look at them on the Merc again.

I have tuned in many stations on the air that were even +50 over and were very clean. And have seen some +20 and dirty.

Bring your Erb amp over here so we can take a close look at its the performance and see if there is any difference from mine.  Or I could bring my Erb to your pad.  

T

[WA1GFZ]

Tom,
I got faked out using my HP141 at a level that generates internal IMD
Yes relative is a good test but not without a step attenuator or something as clean as Mercury.  A quick test if you are generating internal IMD. Drop the input level 10 dB if the IMD drops the same amount you are ok, if IMD drops more than 10 dB the RX your using to measure IMD is generating internal IMD

[WA1GFZ]

My HPSDR

[WA1GFZ]

Modified Erb MRI Amp

[K1JJ]

Nice pics!  So you learned how to upload...

I see the bias board and vertical orientation of the heatsink.  BTW, what was the Richter reading when you got hit over there in pic #2?

Hey did it!!!  Tonight I got the TMC linear working with the HPSDR. 10mW in gives me a VERY SUPER clean 50W pep out. I'm seeing an easy -45 to -50db third IMD now. The S-meter is down around S2 to S3 using the same test as before. The cleanest amp of any yet. I added some more NFB and will try some more tmw. Really pleased.  It's within an S-unit IMD of the HPSDR at 30W pep out. All I need is 30W to drive Dr. Love anyway.

The sweet spot found when the pair of 4CX-250's are running only 650 volts plate and 200V screen with 100ma idle each in class A.  I can barely hear the difference between the -55db HPSDR barefoot and the linear on.   Maybe we can get on with Rob over the next few days and give it an on-air check.  At 50w pep out the plate current does not move at all. The tubes stay luke warm.  Great overkill.

I defy anyone to hear ANY side crud on that signal now...  

I still have a pesky RF intermitant in it. Something drops the power down to 1/4 when it happens.  

T

[w3jn]

Now that I have verified Dr. Love to be clean in the receiver, I have gotten no bad reports.
T

I'll give you a bad report, if you really want one.

[K1JJ]

Just to let everyone feel smug and realize they aren't the only one who screws up big-time...

Yesterday I  dumped 200 watts into my Mercury SDR receiver board and fried the front end. Picked the wrong cable and it became the dummy load... [sigh]

Funny thing is I kept pumping power in trying to figure why it wasn't apperaing on the scope. That board didn't stand a chance. Looks like a fried zener and preamp chip at least.

I was troubleshooting the new TMC amp and started to lose track of things after a long session of looking for an intermittent.

 T

[WA1GFZ]

Tom,
Why not use your HP606 as a source while looking for the amp problem.

[K1JJ]

I WAS using the HP-606.

The SDR coaxial cable was hanging out there from some tests before to see if I had the right amount of drive.  The intermittent was in a coax, so I was rapidly switching them around for tests and finally grabbed the wrong one.

Too bad cuz I had the Mercury BNC pulled off before to prevent such a problem, but it was on this time.

Just like getting cut and going to emergency, I sent it off to Gerd in Germany to repair with his microscope.

T


*** Update: Went to pack up the board and found the box, ESD wrapping, spare Janus and Pinnochio boards. They were covered in cat piss. Looks like "Precious" didn't like the smell of the original user...  Tough week here, electronics-wise.

[WA1GFZ]

now that really sucks
But if I share my week I bet I got you beat
It is even worse when people working for you who don't care blow stuff up.

[K1JJ]

There's even more screw-ups that I didn't mention... too embarassing.  It really comes in trends. The only way is to stick with it until success.

My Merc board is 6-10 days shipping there, a week to repair and the same back.  Maybe a month until I get it back..  My little mistake cost a lot.  Worst case is he says it's toast... :-(

The good news:  That last spike problem on Penny - Phil said he's working on it now and may solve it soon, in software. So I wasn't crazy after all. Still surprised many that had the problem never spoke up. And strange that not everyone had the problem too.

Back on the TMC - I decided to do it right and supply 1500V plate, 350v screen, etc. Might as well make it a stand-alone 200w pep out clean class A amp.  I can easily run it lightly as a driver.  I found even the Henry is up around -45-50db 3rd when run at 50W out.  It pays to run these amps at way reduced power and use more stages if need be..

T

[Mike/W8BAC]

Your not alone Tom. I pumped full output from my Johnson Viking II into the front end of my prized 75A-4 once and held it keyed down wondering why the watt meter wasn't moving. Sometimes the Bird slugs don't make contact internally. The receiver was behind me and it wasn't until I smelled that awful acrid odor of electronic death did I let up. I too misconnected a jumper while testing. That was one of those days I should have seen coming. I had worked for hours ignoring food and rest while chasing a problem. I was tired, hungry and convinced I was close to solving my problem. Instead I made a big mess.

You and Frank are light years ahead of me on many levels and I'm following this thread with interest. Keep up the good work and be assured we all feel your pain. I think I might have an HPSDR transceiver some time soon.

Mike

[WA1GFZ]

Mike,
The HPSDR Hermes transceiver will be out this year. It is Mercury, Penny and Metis all on a 4 by 6 inch board. I think this is a great way to get into SDR.
There were issues driving the atlas motherboard early on but I think solved. All on one board eliminates all the interface issues and extra hardware.

I think I have solved my drop out problems by reducing the buffer sizes.
This allowed me to increase my sample rate to 192k. I was making the mistake of trying to run at the lowest horsepower which limited my performance. I uncapped the headers, rejetted the carb and moved the advance curve up a bit quicker. I didn't hear a single drop out this afternoon.
I was busy working on the big rig so didn't get a chance to tx but hope to get on Sunday. Still only running 16%.

[K1JJ]

Whoops, looks like I missed your post there, Mike.

Ouch on the 75A4.  Yes, I'll bet all of us smoked a valuable something or other at one time or another...   I think the rule is don't work on rigs when we are tired, frustrated or especially ANGRY with our progress. When we're POed, we take stupid risks that we normally would not.

As Frank said, the Hermes sounds like an excellent step forward. It's almost like going from an HT-37/SX101 to a Swan 750 transceiver. (Maybe that's not an upgrade)   The Mercury, Penny and Ozy boards are winners for sure. To combine them all into one board is slick. The only downside is there will not be dual diversity receiving with two Hermes due to technical problems.


Thanks for the encouragement, Mike.   (as we slug on)  I'm working on my TMC class A linear right now.

T

[WA1GFZ]

Tom,
Do you know power SDR has a built in two tone generator? I plan to do some testing Sunday afternoon with the HP141 on the KAZ antenna.