4 x 833A solid state driver

[WA1GFZ]

Tom,
I'm thinking you could have the same issue with phase shift in the PDM filter.
Changing to a two phase PDM would allow the filter corner frequency to raise above the audio spectrum for a flatter phase shift.
Steve is playing with one in his new rig.
Also simulation shows transient response can introduce distrotion. Nothing is cleaner than a class H linear modulator all DC coupled.

[k4kyv]

One of the problems using a backwards-connected push-pull output transformer as a driver is  that the primary winding (now used as a secondary) may not be able to handle the low impedance that the grids present to the transformer over part of the audio cycle. Typically, each one of a pair of class B grids will load the driver stage down to the vicinity of 500 ohms.  Since only one tube is active at a time in class B service, the driver transformer impedance ratio is usually rated to 1/2 secondary.

That means that the total midtapped winding should be rated at no more than 2000 ohms (impedance from midtap to one end of a 2K winding is 500 ohms). Another factor to consider is the DC resistance of the midtapped winding.  It should be no more than about 100Ω from midtap to one side, preferably much less.  A typical 20-watt audio output transformer primary will be rated for anywhere between 5K and 10K, and the DC resistance of each side a couple of hundred ohms. What you need to look for is a higher power output transformer designed for a quad of power tubes in pushpull-parallel.

Before I finally acquired a proper driver transformer to match my quad of 2A3s to class B grids, I used a pair of identical UTC LS series output transformers. One was used with the 2A3 plates connected to the midtapped  primary.  The other was used with the midtapped primary connected to the class B grids. Each transformer was wired so that the midtapped primary served as a tapped inductor. Then I used a pair of oil filled capacitors, about 8-10 mfd each, as coupling capacitors connected directly between the 2A3 plates and the class B grids.  Then I connected the 500Ω output windings of the two transformers together (correctly phased so that the transformer coupling was not bucking the capacitor coupling.  This made a pretty good 1:1 coupling circuit between driver and modulator.

One problem was that the output transformers were not rated for the 80 mills of DC drawn by each side of the quad of 2A3s, and the one on the plate side  ran hot.  The DC resistance was too high for feeding the class-B grids, so I could see some distortion, particularly at the low end of the audio spectrum.  And too, the 1:1 ratio was not ideal for the match.  Even with a quad of 2A3s you need a little step-down.  When I finally acquired the proper audio driver transformer from Dave Schneider, the audio waveform was noticeably improved.

In the interim, I tried a class B driver transformer designed for a single pair of 2A3s.  To reduce overheating at the primary, I tried running my quad at 400 plate volts and 30 MA per tube, instead of the recommended 300 plate volts and 40 MA per tube. This gave the same plate dissipation per tube (12 watts).  The audio with this configuration was superb, since the extra driver plate voltage gave plenty of head room, combined with the step-down of the transformer.  But the BIG problem was that all four 2A3s (good tubes when I installed them) tested weak after only about a year using this circuit. Since putting in the proper transformer and running the tubes at their recommended voltage/current, I have used the same new quad of 2A3s for years and so far they have not deteriorated.

I remember Steve QIX once upon a time before his PDM and class E days, used direct coupling between driver tubes and modulator, employing the midtapped primary of a single audio output transformer. He ran the plates of the drivers at the same voltage as the grids of the modulator, and fed several hundred negative volts to power the cathodes of the driver tubes. The BC-610 uses a similar configuration, but with a regular transformer, in order to use the same power supply for the driver tubes and to provide negative grid bias for the rest of the transmitter.

[K1JJ]

Don,

I must admit, when I take a close listen to your audio, it sounds quite pristine. Out of curiousity I checked your bandwidth the other night and you dropped off big, 5kc away. Very well contained signal. Obviously you have done your homework with the transformers in your system. It's not easy to find the right parts and transformer matches.


Frank -  Yes, maybe I should have moved the corner of the e-rig PDM filter up higher to reduce phase shift. I think you and Jay designed it for about 10kc? or so.  I run the rig at 5kc maximum audio, so maybe it's not an issue.

Steve/QIX is making some good headway with his quad-phase PDM system. "Just like a broadcash station..."


T


"All your base are belong to us."

http://www.youtube.com/watch?v=FVsijmCFs50

http://en.wikipedia.org/wiki/All_your_base_are_belong_to_us

[k4kyv]

Don,

I must admit, when I take a close listen to your audio, it sounds quite pristine. Out of curiousity I checked your bandwidth the other night and you dropped off big, 5kc away. Very well contained signal. Obviously you have done your homework with the transformers in your system. It's not easy to find the right parts and transformer matches.

[/quote]

I can select one of two passive low-pass audio filters in my audio line amplifier, or operate with no filter at all.  Most of the time I use the one that begins to roll off at 5K, with everything gone at 7.5K.  For congested band condx or when there is activity near the frequency, I can switch to the brick-wall filter that cuts off very sharply at 3400~.  With no filter at all, most reports say that the audio still drops off around 7K. Probably the microphone (balanced output taken directly from the crystal element in the D-104) frequency response, since I have run response checks on the transmitter and it is flat out to about 11K.

[WA1GFZ]

Don,
Response is one thing and phase shift over frequency is another.
Tom,
I checked simulation of a number of PDM filters and they all have a lot of phase shift. Mine has over 100 degrees at 10 kHz same with Steve's.
A PDM filter is also an LC delay line so as frequency changes phase changes.
Even if the corner is shifted there is still phase shift. I need to study this some more and think about how it modifies the human voice. I wonder if there needs to phase compensation in the low level audio to make up for this??

[flintstone mop]

You cannot beat a quad of 2A3's for drivers.

The transformer might be hard to find  though
Link to the famous Uncle Raytheon transmitter/PDF

http://www.pa.msu.edu/~edmunds/Tube_FM_Exciter_Transmitter/raytheon_ra_250_am_transmitter.pdf

Even loaded to over 300 watts RF out the modulator still made nice audio and pos peaks to 105%............pair of 810's

[WBear2GCR]

Stefano,

The transformer (for clarity) you and I are discussing is the one that drives the base(s) of the transistors. The only way I know to decide on the value for the resistors on the secondaries is to actually load the transformer to a value that seems about appropriate based on the impedance ratio desired and the transformer's design and then measure it.

Depending on what you find, you may or may not need to add some additional HF compensation.

In audio applications like this (not modulators) I look at the frequency response out past 100kHz AND the square wave response. The square wave response usually "tells the story". Squiggles and overshoot with a 1kHz square wave (that freq usually should look fairly square) can tell you a whole lot.

So, it's a trial and error process to a great extent.

In my experience wideband loop feedback around a transformer can often cause more problems than it solves - it does get the "frequency response" flat, but at the expense of other things... it's likely better to get a better transformer if you need better frequency response than to use wideband loop feedback.

For solid state drivers I do prefer Mosfets... they're making nice high voltage ones these days, and the big advantage is that they tend to clip very nicely compared to bipolars. The bipolars have a very very hard flat top.

                 _-_-bear

[WA1GFZ]

IGBTs are also good for even higher voltage.
It is also stupid to drive transistors with a transformer. The circuit is called a phase splitter. Much cheaper and flatter response.

[Steve - K4HX]

Or switch at a higher freq and the phase shift will be out of the audio range.

Don,
Response is one thing and phase shift over frequency is another.
Tom,
I checked simulation of a number of PDM filters and they all have a lot of phase shift. Mine has over 100 degrees at 10 kHz same with Steve's.
A PDM filter is also an LC delay line so as frequency changes phase changes.
Even if the corner is shifted there is still phase shift. I need to study this some more and think about how it modifies the human voice. I wonder if there needs to phase compensation in the low level audio to make up for this??

[AMLOVER]

Frank,
my project is a B class af amp with 4 x 5868 or TB4-1250 at 4kv plate voltage in parallel push pull arrangement.
What really keeps me back is that I have very little knowledge of solid state technology and limited instrumentation for checking and testing those devices.
On the other side I wish to move further from transformers, mismatches, phase shifting and distortion. In few words if you can provide me your sircuit complete and tested on a board with ability to drive with the necessary af power, voltage, current and bias a final amp with 4 x 5868 then I will be very pleased to get it.
If not then I'll keep all the precious infos of Don, Gito, Bruce, Bear, and Steve and get a Hammond 1650T (1.9K CT to 4-8-16 ohm for 120w) put it back to back with the output transformer of the EL6431 af amp (4xEL34-120w) and go the traditional way as the old memorable boatanchor times having the sweetest sound full of distortion, phase shifting and middle range audio frequencies.
For more infos about my project you can anytime mail me at  info@apsenti.com

Stefano

 

[WA1GFZ]

Stefano,
Stick with what you know. Working without test equipment or basic knowlege of SS sounds like a disaster in the making.
Today I did some PDM filter simulation and tried a wide range of pulse widths into the filter. The delay remains the same for all pulse widths to integrated DC so corner does not seem to be an issue when a pulse is integrated.

[AMLOVER]

Frank,

I think you are very right...I am going to order a Hammond 1650T and I'll follow the old pathway of back to back grids' driving.
I must tell you however that for few moments I was day dreaming a beautiful small board with 2-3 mosfets like the huge ones Steeve discovered lately and some passives with 2 points for audio input 1-2V, two outputs to the final amp's grids, 2 pots for bias controlling (-100v to -140v), one point for feedback connection and 2 points for +-400v power supply.
Day dreaming though can have not always a happy end...
I'll go the traditional way but I'll study all this summer the mosfets and their tips...next Autumn I'll be ready to follow your SS directions and own my SS driver.

Stefano

[K1JJ]

I think you are very right...I am going to order a Hammond 1650T and I'll follow the old pathway of back to back grids' driving.

I must tell you however that for few moments I was day dreaming a beautiful small board with 2-3 mosfets like the huge ones Steeve discovered lately and some passives with 2 points for audio input 1-2V, two outputs to the final amp's grids, 2 pots for bias controlling (-100v to -140v), one point for feedback connection and 2 points for +-400v power supply.
Stefano

Stefano, you have basically described Frank's MOSFET board.  Use the cheap FETS, the 11N90N's as I did - they are only ~$4 each. There's no need for more expensive FETS for audio.

Once you get your Hammond transformer system running, you can slowly put together a MOSFET driver at your leisure. The negative feedback makes a huge difference. I tapped off the modulation transformer secondary since there is only one transformer in line. (The mod xfmr)   It is well worth the effort and you will be plowing new ground for the AMers in your area. 

T

[WD8BIL]

But on a quiet afternoon, at S9+60 over, listening to an SDR sync detector into a big stereo amp system - you will hear the difference.

That leaves out 99.9% of the actual operating conditions!

Getting audio to that level amounts to personal satisfaction in the end. Not that one wasted anytime getting there but for the masses a good SS amp & backwired xformer is more than sufficient.

[K1JJ]

Here's a sample 450 pound capacity lift/hoist on eBay:


http://cgi.ebay.com/Invacare-Hydraulic-Medical-Patient-Transfer-Hoyer-Lift-/200430537475?pt=LH_DefaultDomain_0&hash=item2eaa974b03


General lift listings:

http://shop.ebay.com/i.html?_nkw=patient+transfer&_sacat=0&_odkw=invalid+lift&_osacat=0&_trksid=p3286.c0.m270.l1313


They are patient transfer lifts. The cheapest ones are about $370.   Last year I had one here on rental. I used it to bring in a 450 pound coal stove from the driveway to its pedestal. Did the job myself. These things really work.  

For ham use, the casters may let the lift feet supports slide under the rack cabinet for easy access.

T

You never know who or what you're gonna pick up.

[W2PFY]

JJ I thought the Russian spy was deported? I would have made up secrets to be with her

[WA1GFZ]

looks like a motor hoist.

[AMLOVER]

Tom,

thank you for encouraging me, I'll need only a printed sircuit board.
Somewhere in another thread I think somebody created some boards for this project.
I would go much faster if I can buy one.

Stefano

[W2XR]

JJ I thought the Russian spy was deported? I would have made up secrets to be with her

Terry, I think that was Anna Chapman. She was a hottie, but her heart belonged to Mother Russia.

73,

Bruce