A Simple Test to Check Your Transmitter for IMD (Intermodulation Distortion)

[K1JJ]

Rewriting.

[w9jsw]

Tom,

Did you ever get this re-written? I have an idea...

I run a single 3-500 in the L4B. I built a PS that has a 800VA toroid with 2 650V secondaries in series into a doubler. It puts out 3600V unloaded but sags to 2700. It works ok for 1 tube but the trans is not very stout. Was thinking of measuring the IMD at 2700V, then rewiring the 2 secondaries in parallel to get around 1500V or thereabouts, then measuring. With the secondaries in parallel I bet it would not sag as much and perhaps be a cleaner way to go. Real easy experiment.

John

[K1JJ]

Hi John,

Yes, 3600 to 2700V sag is way too much. That alone will kill your IMD numbers.

Your idea to make a stout 1500V will pay dividends in both HV stability and inherent IMD improvement by running the magic 1500V for the 3-500Z.

If you can live with the reduced power level (90 watts carrier or so) it will make a great "barefoot linear" rig in the class of a Valiant or DX-100 equivalent. The IMD bottle neck will now be your exciter.  By demanding very low power from your exciter, this will help too. (on most rigs but not all)

I use the FT-1000D and single 3-500Z as my new summertime rig.  As a bonus, I am using the same lashup (one 3-500Z at 1500V) to drive my class C  4-1000A plate modulated rig. I simply back off the HV until I get whatever power I need to drive the 4X1 rig. The stress on the FT-1000D is <5 watts now.    I got rid of the Chinese VFO and SS driver mess and the shack is less cluttered.

As far as testing IMD and harmonic distortion, I realized I'd written a lot about it in the various threads about the 4X1s, class A linear chain, etc., in the past. But it's easy enuff for you to inject a 2-tone into your rig and look at how low you can keep the tones when you drive it to your goal power levels. If there is poor IMD you will quickly see it as the 3rd order peaks are no better than -25 DB 3rd.  Read my old threads first.

BTW, as you know the Drake L4B was designed for the input network and pi-network to use two tubes. You may find the swr input higher and the tank circuit not quite optimum for one tube.    The easiest fix might be two tubes and 1500V which would make a SUPER CLEAN 125-150 watts AM carrier.

T

[w8khk]

Power supply voltage sag is indeed a detriment to low values of IMD.  Older power supplies used very high voltage secondaries, with the inherent DCR, resulting in heating and voltage droop.  Large swinging or smoothing inductors, and small capacitors, due to component costs way back then, also compromising power supply regulation.

We now typically use lower voltage, higher current secondaries, with large capacitors, silicon rectifiers with very small voltage drop, and rarely see any series inductors.  But what puzzles me is WHY do we (almost always) use half-wave voltage doublers, when, for a very small incremental cost, we could employ FULL WAVE VOLTAGE DOUBLERS?  See the attached circuit, complements Jim, WD5JKO, (http://amfone.net/Amforum/index.php?topic=45678.msg327555#msg327555)

and the linked article here:

http://www.kwarc.org/bulletin/99-04/tech_corner.htm

For the cost of one extra capacitor, and four more inexpensive diodes, we could double the ripple frequency, and fully utilize the transformer secondary, cutting hum, reducing transformer heating, and significantly improving the output voltage regulation.

This circuit could breathe new life into power supplies like the HP-23, and others in the 800 to 1000 volt range, and with the smaller capacitors available today, they could be used to power clean amplifiers for AM, not just SSB.  

I don't think I will ever build another half-wave voltage doubler anymore!

[w9jsw]

In my case I already had the harbach board. Your idea is much better. Wonder if I can strap on the extra diodes to the existing board? Will take a look.

Tom - I will be driving it with my hermes-lite which has 5W output. May even hook up the feedback loop for Pure Signal. Seems to run fine on one tube though... I am keeping a watch for another. Hate to blow $200 on a china clone.

John

[K1JJ]

Thinking more about your single tube...

The plate impedance may be a wash when you pull out one tube and then reduce the voltage to 1500V.

I just tried two tubes with 1500 volts and set the idle to 130 mA (no bias, true zero bias) and the steady AM carrier drive at 150 watts output.   I loaded the amplifier ridiculouly heavily. (less C2 loading cap mesh)  Barely a hint of color on the plates.  The two-tone test looks great. The 3rd and 5th hardly come off the floor.    The audio peaks seem to have no limit.   Dead carrier eff at about 25%, probably 65% under modulation.  Running the 3-500Zs at about 1/2 dissipation at dead carrier.

I might leave it like this for the hot summer and static.  

Yes, the Hermes and pre-distortion will be a FB exciter. You could run the Variac up to 3KV and still be clean.


T

[KD6VXI]

First things first.

Those are horrible numbers for V Drop from the Antek transformer.  I've used them in SB220 conversions and don't see anywhere near that V Drop.  Have you measured the voltage drop on the primary side?  Running both primaries in phase (if on 120)?  Seriously, that transformer is even more stout than an SB220....  Which makes 1200 PEP with just a couple hundred volts drop.  Something seems amiss to me.

Second, I just purchased a Hermes Lite 2.  What a great little machine!  I really like it..  Was bought second hand from K2SDR, and I couldn't be happier.

I thought about finding a "dots on display" TS440 and using the amplifier as a 100 watt IPA, but then remembered I bought a few 8930s from a member on this site.  Since they are pretty much a voltage drive tube, I could run a pair and get near legal limit (they need appx 2 watts driving power).  However, still stuck with the blower noise.  Decisions, decisions......

Have looked at the LDMOS kits, but still a bit expensive when you factor in all the band filters as well as a tuner.  I wonder, would a link coupled tuner be good for the output of one?  A la Johnson KW level tuner?  Anyone ever measured the filter characteristics of one of those?e

I ran numbers on a 3-500Z running at 1500 volts, 600 watts PEP output.  Plate load is 1875 ohms.

The L4B runs about 2400, and at 1.2k that equates to a 2400 ohm plate load.

The bigger issue is the input Z.  A pair of 500Z tubes is in the neighborhood of 55to 57 ohms.  A single one is 110-115 ohms.  That won't equate to getting proper drive to the tube.

The transformer you're using is the equiv to the original xformer.  A SB220 is appx .6kva.

--Shane
KD6VXI

[w9jsw]

I know, I am not happy with it either. I run it on 245v on the primary. The feed is 25ft of 10ga wire off a 20a breaker at the panel. It is pretty easy to make sure the windings are in phase on these. I don't have enough space to go to the 950V model. So if going to 1500V solves the sag and improves the regulation, I might just add another tube. I bet it would not sag as much if I had left the space heater bleeder resistors in the circuit.

[w9jsw]

Back to the IMD discussion.

As I have a Hermes-lite 2 and a couple of Hermes lite 1's, I think I can do the following.

Use the HL2 with PowerSDR to generate a 2-tone transmit signal. It has that capability in the test tab. I should run that into a dummy load, and use a tap to sample the signal into a second system comprising one of the HL1's showing a panadapter. I have LinHPSDR running on a Raspberry Pi or I could gen up another windows system to run a second PowerSDR. I should be able to read the HL2 barefoot to get a baseline.

Then fire up the L4B into a much larger dummy, calibrate the tap to not overload the HL1 and take a second reading.

Works?

John

[K1JJ]

John,

The potential problem is that the Hermes exciter may dominate the IMD readings. If the Hermes is only -30DB 3rd, then even if the L4B using a single 3-500Z at 1500V is really -46DB 3rd by itself, the Hermes will contaminate the readings giving you something less than -30 DB 3rd.  IE, the readings can only be as good as the worst component (exciter) in the chain.

So unless you use a super clean exciter, you will never really know how clean the 3-500Z is.  I went to the trouble to use the 100 mW FT-1000D tap, into a 1 watt lab amp and into the 4CX-350J linear to produce 50 watts that was a super clean -55DB 3rd driver. It took a lot of effort. Remember the ten 4CX-250 broadband amps used for testing?

I would run your tests and simply expect the addition of the L4B not to deteriorate the overall readings more than a DB.  IE, if the overall readings are nearly as good as the Hermes, then assume the L4B is OK.  The Hermes or whatever exciter you use will set the stage.


Bear in mind that I used the 1500V lower power mode to avoid a lot of blower noise and a cooler shack for the summertime. You could probably run a well regulated 3KV with less drive and approach -40 DB 3rd order - but have a louder blower and more heat in the shack.

Also as a test gear option, you can also use a two-tone generator off the computer - there are several to download free on the web. Also, a good receiver using a 1' antenna off the back and the S-meter will tell you how your 3rd and 5th order peaks are as you tune down the band.  Use a narrow bandwidth on the RX.

T

[w9jsw]

For those that are interested, here are the IMD numbers for the hermes lite 2. Looks like if I keep the power low, the numbers are quite good. -36 ain't bad, however.

[K1JJ]

The Hermes IMD numbers are amazing at lower powers!

I have an older HPSDR, that is not that clean.  

It takes maybe 10-15 watts of carrier to drive my 1500V 3-500Z pair to 150 watts carrier out, so you should be FB.  Run the overall tests and see what the lash-up does now.  Maybe you can eek out an overall -45 DB 3rd IMD driver combo at 150 W out.  If you ever build up a really big linear for ssb, etc.,  what a nice
driver that will make, backed off - not to mention pre-distortion techniques.

T

[KD6VXI]

Use predistortion on the hl2.  Take your sample before the amplifier.

This should get you - 40 to 50 feeding your amp to test

At least up to 20khz wide.  I don't think the predistortion  takes care of anything better.

--Shane
KD6VXI

[w9jsw]

Rewired the trans. Get 1800V idle and 1700V under load. Much better.

Driving it with my yaesu GL-757GXII I can get 125W carrier. I really want to drive it with my Hermes, but the circuit for keying the old amp is 24V. So, whipped up a circuit to take a -PTT signal and close a relay. Ran out of time to play with it. Maybe later.

John

Monday morning edit - Hooked up the Hermes-lite this morning. 5W does not make a mouse fart. Added my 50W linear and at 30W it drives the L4B to around 150W carrier. Will have to hook up the pure signal across both of the units now to insure a really clean input into the 3-500.

On the lookout for a good used second tube. More power is better, right?  

[w9jsw]

Wanted to follow up on this.

I finally bought a second tube. Using this Antek 650V trans with the secondary's paralleled, I get a solid no sag 1800V and it drives the tubes to 800W CW and around 200-225 in AM mode depending on how I set the carrier up in the Hermes.

If I had to do it over again, I would opt for the 800V secondary's and use Rick's FW doubler. This would make a FB power supply for a L4B that came without one.

John

[km6sn]

Where can I buy a Hermes Lite 2?

[w9jsw]

Check group buy portion of this site  - http://www.hermeslite.com/

[KD6VXI]

Hermes lite makes 2 tone IMD measurement easy as pi, too.

As long as you have a second sdr.

What I do is use PreDistortion on the output of the hermes.  Gets the exciter in the - 50s area.  Use that to excite the Amp and thetis has a built in two tone generator (actually just a pair of carriers generated in software).


If I'm using a driver amp (brick or say a tefrode that takes 5 to 7 watts pep, what the Hermes is capable of) nwrap the PreDistortion around the last drive stage.  Then you're still typically in the low - 50s for IMD).

Use second sdr to measure grunge.  A TinySA would work good here as well.  I use my VNWA or SDRPlay in this regard.  The VNWA goes down to minus 90 or better at its noise floor.  The SDRPlay grunge is also below the minus 50 region so works fine for measurement.....  I just question the absolute linearity of the SDRPlay.

--Shane
KD6VXI