Converting Collins 820D-2 AM broadcast transmitter to 4-400As.

[WB6QED]

I have been trying to convert the PA and modulators in my vintage Collins 820D-2 broadcast transmitter from 5-500As to 4-400As and keep running into a brick wall.  No matter what I do, I cannot get symmetrical +/- 100% modulation.  About the best that I can do is -100% and +90%.  At 50% modulation, the envelope is nearly symmetrical.  There is essentially no difference between operation at 250 W output and 1 kW output.

With the 5-500As installed in the transmitter the screen grid and the suppressor grids are tied together.  In the modulator, the suppressor is grounded.  Of course, with the 4-400As installed, there is no suppressor grid to worry about.

It seems like this should be simple plug and play modification, but it hasn't been.  The PA tubes are NOS Eimac and do not show any sign of being emission limited with no change in performance when the filament voltage is reduced or increased.

Collins specifies that the PA load Z to be 3350 ohms.  I have verified this and have tested the transmitter at this PA load Z as well as at 2800 ohms.  It makes no difference in modulation performance.

PA plate voltage is 3000 volts and PA plate current is about 500 mA.  Power output is a little over 1 kW.

PA screen current is about 110 mA and PA grid current is about 50 mA.  These are good numbers for a 5-500A, but a little high for a 4-400A.

With the factory configuration, the grid bias is about -175 volts and will swing positive up to +40 volts.  This results in roughly 50 mA of PA grid current.  I have also tried reducing the PA grid bias to -225 volts which decreases the PA grid current to about 25 mA which is about right for a 4-400A.  Screen current comes down to 80 mA.

I have good symmetrical voltage swing on the plates of the PA tubes.  I can go above 3000 volts and down to zero with a modulating sine wave.

The PA screen is operating at roughly 240 volts without modulation and swings up to about 320 volts and down to 120 volts at 100% plate modulation.  The 820D-2 uses a separate 750 V screen voltage supply with 5 K of series resistance between the screens and the output of the supply.  I tried raising this value to 8K and lowering it 4K and did not see any change in modulation performance.

The modulation transformer has an input impedance of 11600 ohms which is what Eimac states the Effective Load, Plate to plate should be for a 5-500A at 3000V.  The Eimac data sheet for the 4-400A specifies the Plate to plate effective load at 3000V to be 8900 ohms.

I tried substituting a modulation transformer from a Bauer 707 and didn't notice a significant improvement in modulation symmetry.

Does anyone have any suggestions as to what else I can try???  The only thing that I can think to do is to convert the screen supply into a voltage divider off of the high voltage supply in a way that is similiar to that of the Collins 20V-3.

Brian Henry, WB6QED

[KM1H]

If possible try a pair of PL-175A's, there may be something inherent in that extra grid??

Carl

[K3YA]

I would suspect the screen supply.  If I understand your measurements correctly, the average screen voltage drops at 100%plate modulation and the amount of audio modulation on the screen sounds a bit low.  I have always had the best positive modulation of a tetrode final when I derive the screen voltage from the modulated B+, or by modulating the screen supply with a tertiary winding one the mod transformer which provieds nearly 100% modulation of the screen voltage. Like many others I have also used a self modulating screen that employs an audio choke in series with the screen supply, but I don't think that works quite as well.

You could try putting a choke in series with the dropping resistor and see if that provides more audio swing on G2 to benefit your positive modulation.

[w3jn]

With the factory configuration, the grid bias is about -175 volts and will swing positive up to +40 volts.  This results in roughly 50 mA of PA grid current.  I have also tried reducing the PA grid bias to -225 volts which decreases the PA grid current to about 25 mA which is about right for a 4-400A.  Screen current comes down to 80 mA.

-225 is about right for a 4-400 class C, and 80 mA is just about right for screen current.  What are the values of the screen bypass caps?  If the time constant is too long the screen won't self modulate.  Screen voltage seems a bit low also.  You should tune the transmitter for a peak in screen current, rather than the absolute dip in plate current.  The peak in screen current is a bit sharper.  Try increasing the screen voltage to about 500-600V, then adjust the drive for 80 mA screen current.

Take a peek at the GPT-750 manual and see how TMC did it (4-250s in the final, but 4-400s are drop in replacements) http://bama.edebris.com/manuals/military/gpt750/

Another thing you can try is reversing the plate caps on the modulators, see if the peaks improve.

[flintstone mop]

If memory serves..........."No Money Mike" Dorrough has several of these fine transmitters and changing freqs is not very easy. Are you sure all of the alignments have been done properly??
He mentioned the need for some out of the ordinary test equipment to achieve good results.
Fred

[WD5JKO]

I have good symmetrical voltage swing on the plates of the PA tubes.  I can go above 3000 volts and down to zero with a modulating sine wave.

  Brian,

  In order to verify what you quoted above, you need an AC compensated 1000X scope probe capable of withstanding 10KV or so. Not many hams have such a thing. So if you have one, I'd overlay the modulated plate voltage on one scope channel, and the RF PA screen voltage on the other. Also look at the RF G1 voltage.

  Many rigs on the air today operate 'as normal' with characteristics similar or worse than yours. Upward modulation peak power capability is influenced by many factors including, power supply regulation, power supply source impedance versus audio modulation frequency, RF PA grid bias peak voltage, grid current, RF PA loading, line voltage regulation, etc. etc.

  Does the rig behave the same at 300hz, 1000hz, 3000 hz?

  Try to pin down a variable that makes matters worse since making it better has been harder to pinpoint.

  Maybe the rig behaved the same way with 5-500's??

If the screen voltage maximum is the issue, consider a capacitive voltage divider from the plate supply with a large series resistance. The idea is to make the screen stay in phase with the plate, and do so up to the highest audio frequency transmitted. The screen bypass capacitor value can be an issue if not compensated for.

Edit: If the problem is with modulation linearity, there will be a 'kink' in the Trapezoid pattern. If the Trapezoid is linear (no slope changes), then the issue is with the modulator.

Jim
WD5JKO

[WB6QED]

I appreciate all of the thoughts and insight!  Yesterday morning, I inserted a 8 H choke in series with the screen voltage dropping resistor and significantly improved the modulation performance.  With the 8 H choke installed, I was able to get nearly symmetrical modulation with less than 1% THD+N at 90% modulation at 100 Hz and 1 kHz.  The THD+N rises to a little over 3% at 10 kHz on high power (1 kW), but is around 1% on low power (250 W).  Any thoughts on what the cause of this might be?

It isn't quite perfect yet, but it is getting there!

When the 5-500As are installed, the screen and suppressor grids are tied together.  Perhaps that helps them to self modulate better?  It is sort of surprising that Collins chose to use a screen voltage power supply with a choke input and capacitive output.  It seems like it might have worked better the other way around.

I don't know how the transmitter performed with 5-500As installed.  When this one came into my possession it had already been modified (sort of) for use with 4-400As.

I have a Tektronix P6015 X1000 HV probe so making measurments on the modulator isn't a problem.  I like the idea of viewing the phase relationship of the plate modulation and the screen modulation.

As part of the troubleshooting process, I did a low voltage test of the modulation transformer.  10 VAC into the primary results in 8 VAC on the secondary.  1.25:1.  The output voltage was symmetrical from each side of the primary.

The screen bypass capacitors are 1000 pF.

I think that it is preferable to use the separate screen voltage power supply rather than a voltage divider off of the modulated HV to perserve good performance at low power.

Using an HP 4815A Vector Impedance Meter, I have measured the PA load Z to be:

1480 kHz:  2800 ohms + 10 degrees
1490 kHz:  3200 ohms + 4 degrees
1500 kHz:  3600 ohms -2 degrees.

I have taken the tracking generator from a Tektronix 7L5 and gone in backwards and looked at the output filter bandpass symmetry at the PA plate and it is centered and symmetrical on each side the carrier frequency.

Thanks again for all the helpful ideas and insight! 

After I have the 4-400A conversion sorted out and the audio performance within original specifications, the next challenge will be to make the leap to 75 meters!

Brian, WB6QED

[K5UJ]

I don't know how the transmitter performed with 5-500As installed.  When this one came into my possession it had already been modified (sort of) for use with 4-400As.

I have been told by an 820D owner that the stock rig originally could only hit as high as 110% positive.  Later, Collins offered a different mod. transformer that got it to go up to around 120 or 125%.   I think that tx came out when everyone wanted to be LOUD so there was dissatisfaction with the original lower mod. limit.   It's a beautiful rig.  The one I have seen has the 5-500s on the right side of the cabinet facing each other so it must be an earlier version.  The build quality is wonderful to me.

Rob

[WB6QED]

The 820D-x version with the 5-500As on the right side of the cabinet is the 820D-1.  I believe that the modulation transformer that is installed in the 820D-2 has the capability to do +125% positive modulation.  Interestingly, the factory test proceedure for the 820D-2 specifies that the static bias for both 5-500A modulator tubes can be as high as 300 mA.  That is almost twice as much as what I have seen with other plate modulated broadcast transmitters.  When the bias is set that high, the plates glow very brightly!

Brian, WB6QED

[KA3EKH]

Got a schematic of the transmitter? Including the modifications that were done to it?  Looking to pick up an old "Rock" over in NJ and helping a friend with converting it so interested in seeing what's going on. The only Collins Rock I have worked on in broadcast service used all 4-400  I thought, just looking at the front of the transmitter kind of surprised that it was not a PDM modulated transmitter being that almost all other transmitters that look to be that age were PDM by then but have to be careful, already upset the SDR people this month don’t want to start something with the Collins people thinking they may be worse!
Ray F.

[flintstone mop]

The 820D must have been in the same class as the McMartin BA1-K, which used 5-500's also. And the designer cabinet fit for any living room. Very simple straight forward design and plenty of modulation. The weird thing was the divide-down circuit used to get to the desired on-air freq. in the osc.circuit.
I bypassed that mystery circuit and provided grid drive with an antenna tuner and my mobile TS 50.

[KA3EKH]

I was fortunate in only having to work with a McMartin once. Have to love the seventies style and the real wood grain contact paper and the little rectangle plastic buttons on it kind of like their EAS box, also recall that it had a bad habit of tripping off air when you selected different meter positions for the test meter. But that was a FM transmitter, never worked with one of their AM transmitters.

[WB6QED]

Got a schematic of the transmitter? Including the modifications that were done to it?  Looking to pick up an old "Rock" over in NJ and helping a friend with converting it so interested in seeing what's going on. The only Collins Rock I have worked on in broadcast service used all 4-400 

I haven't scanned the schematic or the manual yet.  The schematic is a massive single sheet.  At some point, I will try to figure out a way to do it.  The 820D-2 is solid state all the way to the finals and modulators.  So far, I have only changed the filament transformers from 10 VAC output to 5 VAC output.  Peter Dahl made a couple for me that were exact physical replacements that look original.  The other soon to come modification will be the addition of an 8H choke at the input to the screen voltage dropping resistor.  I think that I can use standoffs to install the new one on top of the existing one and have it look very nice.  This is still a work in progress with the ultimate goal of moving it up to 75 meters.

To my knowledge, neither the 820D-1 nor the 820D-2 came from the factory with 4-400s installed.  So, the one in New Jersey must have be converted.  My guess is that they changed the filament transfomers but didn't make any other changes when they did the conversion.  The 4-400As use the same pin out as the 5-500As.

Collins discontinued the 20V-3 and introduced the 820D-1 in 1968.  I believe the 820D-2, "The Rock," was introduced in 1974.  It remained in production until the pulse width modulated 828C-1, "The Power Rock One," (often referred to as the "Power Pebble") superceded it around 1979.  Shortly after the 828C-1 was introduced, Continental Electronics purchased the broadcast division of Rockwell/Collins and changed the model number to 314R-1.  I believe that it remained in production into the mid-1980s.  It is the last AM broadcast transmitter design that Collins produced.  To my knowledge, it is the only 1 kW pulse width modulated AM broadcast transmitter that used electron tubes and in more ways than one represents the end of an era.

Brian, WB6QED

[WB6QED]

The 820D must have been in the same class as the McMartin BA1-K, which used 5-500's also.

The schematic that I have shows that the McMartin BA-1K used 4-500s in the PA and modulator so perhaps there were different versions?  As with the Collins 820D-2, it uses a separate screen voltage power supply for the PA screens.  It has a different circuit topology though.  It has a 10 uF capacitive input with a 10 H choke in series on the output that goes directly to the screen voltage dropping resistor.

Brian, WB6QED

[KA3EKH]

Must be my mistake, the only Rock I have seen was at a transmitter site for Hell hole of a station in Ocean City Maryland on 1590, at the time it was removed from service and they were running a beat Harris SX and a Sparta as a backup. The Collins was disconnected and pushed aside at that time, struck me odd they wanted a Sparta as a backup when they had the newer looking Collins but have been told that the Sparta that appears to be from 1950 is actually a newer transmitter that was built from a old design. So assumed there were 4-400 in the Collins and did not know they were 5-500, the transmitter over in NJ I have not seen yet.
Ray F.

[WB6QED]

When they are working correctly, the Harris SX series transmitters have spectacular audio and RF performance (I have worked on many of them!).  Sparta purchased the Bauer transmitter line from Granger in the 1970s.  The Sparta transmitter was probably an updated version of the Bauer 707 which was introduced about 1960 when the FCC authorized Class IV AM stations to operate at 1 kW during daytime hours.  The Bauer 707 could be purchased as a kit.  Paul Gregg told me that during one of the NAB shows in the 1960s, they even built one up from scratch during the convention.

Maybe they had the Collins disconnected because they couldn't find tubes for it???

Brian, WB6QED

[KA3EKH]

This was one of those little signal owner no money operations and from what I recall the SX had a weird combination of new and old output cards and the Bauer looked very old school with all octal tubes and plug in relays on the front, I can see that as a kit because looked like all you needed to build one was a screwdriver and a hammer although think the owners only owned a pair of worn out channel locks and a penknife. I can be hard on those people because I was filling in for a friend the couple times I went there.

[N0WEK]

Got a schematic of the transmitter? Including the modifications that were done to it?  Looking to pick up an old "Rock" over in NJ and helping a friend with converting it so interested in seeing what's going on. The only Collins Rock I have worked on in broadcast service used all 4-400 

I haven't scanned the schematic or the manual yet.  The schematic is a massive single sheet.  At some point, I will try to figure out a way to do it. 
Brian, WB6QED

I got the 3 ft x 5 ft schematic for my Gates BC-1J scanned at Kinko's along with the whole manual. I had three manual copies, a couple of copies of the large schematic and an electronic copy on a thumb drive for a total of about $100.00.

The 1955 blueprint was in bad shape and the original scan was really dark and low contrast. Kinko's cleaned up the file and it now prints out fine.

Kinko's scanners will copy up to 42" wide and unlimited length.

[k4kyv]

Have you tried running the 5-500As as real pentodes instead of as pseudo tetrodes?  With a pentode tube, the suppressor grid is supposed to run at zero volts or at least at lower voltage on the suppressor than on the screen. None of the tube data I can find in a quick search says anything about the suppressor voltage. I would try grounding the suppressor and seeing what happens. Or look up the data on several other transmitting pentodes, like the 803 and 4E27, to get a general idea of typical grid voltages, and scale the plate/screen/suppressor voltages roughly proportionally.

[KM1H]

The 5-500A suppressor is designed to be grounded and is connected to the metal shell so I dont see how it can be connected to the screen.  Maybe the shell is floating in that TX.

Carl

[WB6QED]

Here's the latest update. 

I was able to improve the modulation symmetry by adding an 8H choke on the output of the screen voltage power supply at the input to the screen voltage dropping resistor.

Last night, I took a quick look at the audio output spectrum coming out of the high side of the modulation transformer using a Tektronix P6015 HV probe and a DC block to AC couple the signal into my 7L5 low frequency spectrum analyzer.  With the 10 kHz audio input level adjusted to roughly that which corresponds to 100% modulation, there is a considerable amount of even and odd order harmonic distortion.  The second harmonic is only down 30 dB from the fundamental with the rest of them gradually falling off.  All of this harmonic energy is translated to the RF spectrum when the PA is modulated.  This also occurs at lower audio frequencies, but it isn't quite as severe.

When the audio input level is reduced, the harmonic energy starts falling off rather quickly.  A 10 dB reduction in input level nearly eliminates the harmonic distortion.  The high level waveform isn't clipped.  It seems as though something is being driven into overload, though.  I was suprised to see so much harmonic distortion at the output of the modulation transformer.  Having never looked at the audio waveform at this point before, I would have expected the harmonics to be down 50 dB or more when the modulator section is running at full power (in this case on low power).  Has anyone ever made a measurement at this point before?  What should I expect?

Is there easy way to determine what the proper load impedance for testing the modulation transformer is?  I am not certain that the test I made was valid with the modulator still connected to the PA.  I am beginning to think that I need to isolate the modulator from the PA to verify/troubleshoot the problem.

Of course, there are any number of things that can cause this problem but perhaps there are some thoughts as to where to begin?

Thanks for everyone's help!

Brian, WB6QED

[WB6QED]

The 5-500A suppressor is designed to be grounded and is connected to the metal shell so I dont see how it can be connected to the screen.  Maybe the shell is floating in that TX.

The PA tubes are mounted on a piece of green fiberglass that floats the shell of the tube.  The suppressor grids are then tied to the screens.

For the modulators, the suppressor grids are tied to ground.

Brian, WB6QED

[KA3EKH]

Was just talking to someone about this and have to ask, is this the transmitter where the cathode of the PA tube is run at a high B- potential and the plate is at ground potential? Are they doing the same thing with the modulators? Worked with things like the Harris and Continental series modulated PDM transmitter but never heard of such a thing with an old design plate modulated system then again that stuff was built back in the sixties and there were lots of good drugs around then. Want to see the schematic more than ever now.
Ray F.

[k4kyv]

The 5-500A suppressor is designed to be grounded and is connected to the metal shell so I dont see how it can be connected to the screen.  Maybe the shell is floating in that TX.

The PA tubes are mounted on a piece of green fiberglass that floats the shell of the tube.  The suppressor grids are then tied to the screens.

For the modulators, the suppressor grids are tied to ground.

Brian, WB6QED

I got it backwards.  You are trying to go from 5-500As to 4-400As, not the other way round.  But I still don't understand why they tied the suppressors to the screen instead of to ground.

I would guess your problem is getting the correct ratio of screen modulation to plate modulation, which is apparently different from that of the 500s. I assume the 4-400s are known to be good.

This is exactly why I prefer to use triodes for anything over 100 watts.  No hassle with the screen supply, screen modulation and screen protection circuitry.  The extra grid drive and neutralisation requirements that go with triodes are less trouble. Of course, I am aware that you are trying to make an existing transmitter work, not build a new one.

I would think any circuit that works with 4-400s and gives linear modulation (as checked with the trapezoid scope pattern) with headroom beyond 100% positive, for example in the T-368, KW-1, GPT-750, 20V, Johnson KW, BTA-1, etc., all of which are widely used on the ham bands, could be reproduced to work equally well in this transmitter. If the 4-400s can be tested in another transmitter and proved to modulate satisfactorily, then there must be some parameter in your circuit that is not correct. It may not have modulated satisfactorily even with the original pentode tubes.

I have had a similar problem with HF-300s, even ones that check good, including a couple that are supposed to have been brand new and unused.  The trapezoid shows curvature on the sides of the triangle, whereas it should be perfectly straight.  It also causes the plate current flicker downwards at a greater percentage than the PA plate voltage, as a result of the negative carrier shift. Some tubes show more non-linearity than others, but of all the tubes in my collection, only one or two sets modulate perfectly linearly.

The pair I am using right now will modulate linearly at reduced plate loading, but at full rated plate current, non-linearity begins to show up, so I have been running them lightly loaded.

I would suggest connecting up the scope for a trapezoid pattern, using a sample of the rf output to deflect the vertical plates, and audio from a R-C sampling network directly off the secondary of the modulation transformer to deflect the horizontal plates.  Note the non-linearity. Is the curvature uniform, from zero percent modulation to 100+ percent, or does most of the curvature occur around some specific percentage of modulation?

With the trapezoid, you can  look at the base line to check the percentage of modulation as seen directly off the modulated +HV line, in other words, the DC input to the final. The modulated RF percentage, as indicated by the vertical deflection, should be identical to that of the base line, and the sloping sides of the triangle a straight line.  If the modulation percentage as indicated by the horizontal vs vertical deflections are different, you will see a curvature on the sloping sides of the triangle, indicating non-linearity of the modulation at the final.

[WD5JKO]

Is there easy way to determine what the proper load impedance for testing the modulation transformer is?  I am not certain that the test I made was valid with the modulator still connected to the PA.  I am beginning to think that I need to isolate the modulator from the PA to verify/troubleshoot the problem.

  Brian,

   Your thinking is spot on. Separating the two systems is desirable, but difficult. Sure, you can make the modulator load into a resistor who's value is R=RF Final plate voltage/RF Final plate current.  If the screen current comes from the plate supply, then add that in as well.

  The problem with the above idea is that the DC magnetizing current (Final plate current) in the mod transformer is eliminated, so the modulator will work better than with the RF PA attached. If the system uses a modulation reactor, then this is not the case. Also if you load with a resistor, grounding the secondary winding will put stress on the primary to secondary insulation, so you may want to keep the resistor load at the same potential as the modulator plate voltage. Don't float the scope however.

   If this rig were on my bench, I'd take a different approach. I would configure transmitter to be able to get transmitter a trapezoid pattern on the scope, like what Don said. A compensated voltage divider is needed (could use your 1000X HV probe) for X axis deflection (modulated B+), and the RF carrier attenuated for the scope Y axis, and use XY mode on scope. Here you will see any modulation non-linearity as seen as a slope change on the trapezoid walls.

   The neat thing about the trapezoid is that it is not influenced by modulator overload or distortion. So work on the RF PA until you can modulate linearly to 100% or higher. Then switch to envelope pattern to look at modulator distortion. A triangle wave, mid band at say 1 Khz, should be perfect looking. Any slope change as level increases indicates modulator linearity problems. Doing the audio triangle test when the RF modulation linearity is bad is a total waste of time. Fix the RF system first, then the audio.

   I can do distortion tests here using a PC based FFT software program, and a good sound card with suitable I/O interfacing.

http://www.dr-jordan-design.de/Download/examples_PA_futuretec.pdf
(Educational version available for about $50 shipped)

   Here I built a low distortion RF detector designed for a level of ~ 3 watts AM. Basically a diode detector followed by an emitter follower with +/- supplies. The detector needs a lot of attention, and THD can be made very low. With my Retro-75 transmitter (modified), I can get THD+N < 0.5%  up to 80% modulation. Here my keep alive supply limits the negative peaks to ~ 85%, so a soft negative clip increases THD rapidly such that the positive peak at 120%, and negative at 85% gives ~ 20% THD. With asymmetrical mic audio of proper polarity, the negative peak limiter is only occasionally working.

  So building a low distortion detector scaled for the power level you run might be a worthwhile task if you want to make distortion tests of the transmitter that include both modulator and RF PA.

Jim
WD5JKO