New 4X1 X 4X1's Rig - Problem with audio spikes - Questions - Pictures

[K1JJ]

Hola!

Well, finally worked out the many problems this 4-1000A modulated by a pair had in store for me. I was so confident it would work right from the start, but it took several days to work them out.

Here's the original thread for reference:     http://amfone.net/Amforum/index.php?topic=21969.0

I swept it and the audio looks quite good. It may be one of the better performing plate modulated rigs I've built so far.  Originally the blower was mounted on the back of the sub-chassis. The noise was terrible!  I mounted the blower on the outside house wall. It's now quiet and the air is cool enuff so I can run the blower even slower...  Good air flow through the tube seals and chimneys.

Anyway, here's the problem...  When I unkey the modulators, I get a big audio spike that makes the modulator plate meter jump almost 1A and the mod xfmr gaps spark. I see the spike come back on the audio line into the low level stages too.  If I disconnect the 200W solid state amp output, (which drives a backwards connected 8 ohm audio xfmr) the spike is gone - so it's not the modulator tubes themselves.

I tried keeping the mod tubes keyed all the time and then unkey the final and HV, but it's still there. I tried keeping the HV on the rig all the time, but same thing.

I put 800 ohm resistors as a test into the modulator grid leads, but still had a bad spike on unkey.  I have a delay in the ant relay that keeps it on for 500ms to dump the RF power on unkey, and tried changing it to instantaneous unkey - same spike problem.

When the audio input to the SS amp is disconnected, no change.  I put .001's to ground on the output leads of the driver xfmr secondary - NG.  I may try some on the 8 ohms leads - Haven't figured what values to try there yet.

It appears the spike is getting generated within the SS amp itself during the RF unkey. It's in a well shielded cabinet. I've used it many times in the past for this purpose with no problems.

The rig sounds so good in the monitor I'm dying to get it on the air, but don't dare cuz of the risk of blowing out my mod iron. (1KW RCA BC piece)

Any ideas?    If I can't get this working right I may have to build up a tube driver maybe using 813's, but I don't want all the heat just for a driver...


*** Here's some shots of Fabio. Notice the ducting to the wall. It leads to a squirrel cage outside pushing air into the rig. I figure the air is cooler than inside most of the year, so why not?  Pictured are two modulators and one RF final on the same chassis. It is neutralized, audio negative feedback around the modulators, vac variables, bandswitched, 70H in the Heising, 20H screen choke, regulated 90V fixed bias with 125V grid leak, variac on screen supply, 3500V on final, Plexiglass view and covered fully for safety.


T

[K1JJ]

More shots with Plexiglas removed.  All parts are easily accessible for servicing.

[K1JJ]

#1 shows the problem SS amp on top:

#2 below is with the full dress Plexiglass protection:

[W2XR]

Hi Tom,

She looks like a beauty!

Have you considered the transient to be possibly orginating within the negative feedback loop between the plates of the modulator tubes and the solid-state audio driver stage? Have you disconnected the loop and observed the results? Is it possible the high-voltage DC blocking caps in the feedback loop are discharging upon key-up and creating the transient in question here?

I'm looking forward to hearing that rig, and working you, of course. I put my rig back on the air last weekend, and I was on for about an hour-and-a-half this afternoon.

73,

Bruce

[kc6mcw]

Very nice Tom!

[K1JJ]

Hi Bruce,

Yes, I disconnected the NFB as a first test.  No effect. I can disconnect the input to the SS amp (with the NFB) and it still generates a huge spike on unkey.  The only way to make it stop is to either turn the power off on the SS amp or unplug its output from the driver xfmr.  I can also see the modulator tubes flash a faint blue color as the spike occurs. There was also some arcing within the modulator components that I fixed - shows me the voltages generated are way too high for longevity and this problem must be fixed.


It will be nice to hear your big 4-400 rig and join in the plate modulated scene again...

Thanks for the comment - well, I KNOW it's not in a all-metal paneled rack and it could be made prettier. But it will function flawlessly when I'm done - that's for sure. I do have a few amps that are in all metal cabinets. But this rig is a showboat that I just love to look at every day and see it operate like the invisible motor or invisible man toys...  The enclosed metal rigs are FB, but I soon forget what's inside...  This keeps it magical for me.

T

[W2XR]

Hi Bruce,

Yes, I disconnected the NFB as a first test.  No effect. I can disconnect the input to the SS amp (with the NFB) and it still generates a huge spike on unkey.  The only way to make it stop is to either turn the power off on the SS amp or unplug its output from the driver xfmr.


It will be nice to hear your big 4-400 rig and join in the plate modulated scene again...

Thanks for the comment - well, I KNOW it's not in a all-metal paneled rack and it could be made prettier. But it will function flawlessly when I'm done - that's for sure. I do have a few amps that are in all metal cabinets. But this rig is a showboat that I just love to look at every day and see it operate like the visible motor or visible body toys...  The enclosed metal rigs are FB, but I soon forget what's inside...  This keeps it magical for me.

T

Hi Tom,

Yes, it does look like the Renwal of ham radio transmitters! Renwal was the maker of the Invisible Man/Woman and V-8 engine kits!

What happens when you leave the entire rig intact, but short the audio input only to the SS audio driver, and then key-up?

For what it is worth, in my rig I apply the 1250 VDC to the 845 audio driver last during the transmit start sequence. I also interlock the 1250 VDC audio driver B+ so it turns off first during the transmit-to-receive sequence. Note that I did not do this for purposes of elimination of any kind of key-up transient, but to ensure that I can't apply audio from the 833A modulator to an essentially turned-off final amplifier stage; the modulator would be unloaded in that case.

Keep feeding us ideas and I'm sure we can determine the origin of this transient.

73,

Bruce

[N2DTS]

Well, keep in mind plexiglas is no sort of shield, there will be lots of rf in the area.
That might be getting into the solid state amp.
Where is the amp located when you were doing the testing?

My 3x4D32 rig had spikes on unkey, not on the mod iron but an RF spike, and talking with Chuck one day
(wb3fjj) he said he had a rig do that and it was parasitics, i added neutralization and they went away...

Maybe some odd length of cable between the rig and the audio amp, or its unshielded...

My amp is in a cabinet on the desk, and shielded cables run to various modulators.
I used some twinax stuff, twisted shielded cable, about 14 gauge wire inside.

I heard you testing today, you were about 10 over in South Jersey.
Something did not sound quite right with the audio I think.

Brett

[N2DTS]

Looking at the pictures, the mod and rf tubes are right next to each other?
How do you keep rf out of the audio?
I try and keep the modulators in their own cabinet, and the rf in the other, even at lower power levels I run.

Brett

[K1JJ]

OK Bruce and Brett -

Bruce:  I will try your suggestion of shorting the SS input and keying up. Though I did try turning its low level pot all the way down and it helped a little, but not much.

I like your idea of sequencing the unkey audio, in my case.  Maybe I can unkey the SS amp output first, then unkey the rig. That is a pain though, vs: making it a delay.

Brett:  You heard me testing ??!!  I was testing into a Bird dummy load this evening located in the shack.  But maybe there was some leakage into the antennas, cuz it uses a common switch.  What frequency did you hear me on?  That will tell the story. But yes, the audio was horrible when I first fired up. It progressively got better until I thought it sounded as good as any plate modulated rig I've built so far. It's quiet running and has some soup, so I'm excited.


Yes, all audio leads are shielded in and out. The SS amp sits up on the rack underneath the steel cabinet rack top. (see the pic) The RF key and unkey looks clean on the scope - it's just that local mod iron spark gap arcing and current meter 1A jump that is troubling.

T

[K1JJ]

Looking at the pictures, the mod and rf tubes are right next to each other?
How do you keep rf out of the audio?
I try and keep the modulators in their own cabinet, and the rf in the other, even at lower power levels I run.

Brett

Broadcash rigs mount their mod and RF tubes together all the time. Nice for air flow simplicity. I see a few 4-400 models that have them in a quad line up.  The mod tubes are high level at that point, so a gain of 10db in triode connected 4X1 service is no problem.  I see no signs of RF in the audio during normal operation.

However, your observation may uncover a clue as to why this spike is occuring. Must think on it.

Remember that is does not happen if I discoonect the SS amp output or turn off the SS amp.  It appears to be generated within the SS amp itself. I'll bet a different amp or tube driver wud solve it, but I don't have either available right now. This amp has worked b4 with 4X1 X 833A's in Plexiglas as well as other exposed rigs like this in the past.  Maybe a focus on SS amp filtering in/out is in order.

T

[N2DTS]

It was in the morning, or around noon time I think.

If you get the spike with the amp on, but not with it off, its rf in the amp.
Cant see it being anything else since turning the amp off eliminates it...

Brett

[N2DTS]

Broadcast rigs only run at lower frequencies.
lots of RF around the modulators, and back into the driver trans into the amp?

Brett

[WA1GFZ]

Tom,
You can store a lot of energy in a 70H choke and mod iron primary. Maybe you need to bias everything off before you remove the load. I would kill the RF drive and remove audio drive first, Then remove the antenna. Just having DC current flowing through the modulation transformer can induce a big spike if it turns off quickly. You might want to bias the modulators into cutoff before the relays drop out. Just like in an old point ignition, you get the spark when the points open and primary current stops flowing..

[K1JJ]

Brett - You heard the testing in the afternoon and morning - good. I was testing only in the evening after dark on about 3856 into the dummy.

Yes, no doubt the RF is being generated in the SS driver amp on unkey only.  I'm just trying to come up with a way to fix it.

I'll try some more filtering ideas tmw and think more about a relay sequencing during unkey too.

T

[K1JJ]

Tom,
You can store a lot of energy in a 70H choke and mod iron primary. Maybe you need to bias everything off before you remove the load. I would kill the RF drive and remove audio drive first, Then remove the antenna. Just having DC current flowing through the modulation transformer can induce a big spike if it turns off quickly. You might want to bias the modulators into cutoff before the relays drop out. Just like in an old point ignition, you get the spark when the points open and primary current stops flowing..

Good thinking, Frank.

Yes, there IS huge energy in that 70H inductor during unkey. Where does it normally go?    I always thought that keeping the RF final TR relay on as the last relay was to dump that energy into the final/antenna.  I know sequencing can get complex.

I'll have to think more about first biasing off the modulators on unkey. I could switch in some diodes in the fil CT to do that. But still, that inductor energy needs to be dumped somewhere - into the final as I described above?  Right now the TR antenna relay unclicks 1/2 second after all else is unkeyed. That's the only sequencing I have right now.  The RF carrier unkeys very cleanly.

T

[W2XR]

Tom,
You can store a lot of energy in a 70H choke and mod iron primary. Maybe you need to bias everything off before you remove the load. I would kill the RF drive and remove audio drive first, Then remove the antenna. Just having DC current flowing through the modulation transformer can induce a big spike if it turns off quickly. You might want to bias the modulators into cutoff before the relays drop out. Just like in an old point ignition, you get the spark when the points open and primary current stops flowing..

Tom and Frank,

As goes without saying, the start-up/shut-down sequence in any high power plate modulated transmitter is very critical.

I have 40 hys in the mod reactor, a Gates BC-1G mod xfmr, and a 2 uf DC blocking capacitor, along with 38 uf of distributed energy storage/filtering/decoupling in the HV PSU section my rig (probably roughly equivalent to Tom's rig), and I do not have any kind of turn-on/turn-off transients of any significance anywhere in the transmitter. This is because the transmitter turn-on/turn-off sequence largely prevents this from happening, by virtue of interlocked time-delay relay logic switching of the HPA RF drive, HV, HPA screen supply, audio drive, and T/R relay switching. The entire start-up time takes about one second, with the bulk of this delay (about 750 ms) in the HV soft-start. The shut-down sequence takes about 750 ms. I never claimed that my rig was ideal for fast break-in operation; it was'nt designed that way!

I'm sure that you have largely completed the electrical design and construction of the rig, and it may well be a PITA to add the appropriate time-delay relays and circuitry, but the protection and reliability improvements are certainly worth the time and cost involved.

Just my 2 cents!

73,

Bruce

[K1JJ]

Sounds like a nice sequencing setup, Bruce.

Yes, I plan to add a system here, now that I see it may solve the problem.

For the Class E rig I used a sequencing board that allowed up to six events to be programmed. Worked out FB.

I can eliminate the HV 750 ms delay, cuz it's on all the time - I just key a relay to direct the HV to the particular rig I'm using - quick key up.

So Bruce, how would you suggest I sequence the various events, A-Z?    I've done it all b4, but is hazy at this point for a plate modulated rig.  Bear in mind the modulators are triode connected with the driver xfmr CT to ground, so no grid or screen bias to deal with there.

T

[WA1GFZ]

Tom You could just open the CT with a relay contact with a large resistor across the contacts. Also put a resistor in series with cap across the contact so it shuts down slowly.

[W2XR]

Sounds like a nice sequencing setup, Bruce.

Yes, I plan to add a system here, now that I see it may solve the problem.

For the Class E rig I used a sequencing board that allowed up to six events to be programmed. Worked out FB.

I can eliminate the HV delay, cuz it's on all the time - I just key a relay to direct the HV to the particular rig I'm using - quick key up.

So Bruce, how would you suggest I sequence the various events, A-Z?    I've done it all b4, but is hazy at this point for a plate modulated rig.  Bear in mind the modulators are triode connected with the driver xfmr CT to ground, so no grid or screen bias to deal with.

T

Hi Tom,

Glad to be of help! Here is how I do it in my 2x 4-400A by 2x 833A rig.

Here goes:

A) For transmit start-up, the following events occur serially:

1) Plate contactor closes 220 VAC circuit to HV/plate PSU

2) Soft-start limit resistor is shorted approx. 700ms. later.

3) Apply HPA screen voltage

4) T/R relay contacts transfer

5) Apply RF drive to HPA

6) 1250 VDC audio driver plate supply contactor closes, enabling audio to be applied to the modulator stage.

B) For transmit shut-down to receive mode, the following events occur serially:

1) 1250 VDC audio driver plate supply contactor opens, effectively removing audio drive to the modulator stage, precluding the possibility of the modulator looking into an unloaded condition.

2) RF drive is removed from the HPA

3) HPA screen voltage is removed

4) 220 VAC plate supply contactor opens, removing the plate voltage from both the 4-400As and the 833As.

5) T/R relay contacts transfer to "receive".

The time delay relays I use are made by a company named NCC (National Controls Corp.). They are microprocessor controlled and programmable by virtue of DIP switch selection, and plug directly into a standard octal socket. Hell, you can even DIN-rail mount them if you want for housekeeping purposes, or if chassis real estate is now at a premium. I bought mine years ago thru e-Bay, but both Allied and Newark carry them, and they are reasonably priced. I bypass the 115 VAC input to each TD relay with a couple of .001 uf caps to ground, and never have any issues with RFI affecting their operation, etc.

Forget about those adjustable RC time constant-controlled TD relays; the ones with the control knob on the top. They are fine for use in a soft-start circuit for power tube filaments, etc., but not in those applications where you want precise and repeatable time delays. This is why I used the NCC relays.

Let me know if you need more help with this!

73,

Bruce

[Gito]

Hi
 
Maybe a stupid idea from me,
unkeying such big transmitter can cause a spike in the power line/220 VAC line.
This spike got into the SS power supply trough the AC line,maybe small but amplified by the SS Amplifier and drove the Modulator,and since its not loaded( the RF amplifier is of) than a big surge current gets into  the modulator,makes a spark jump across the Gap.
Try to use a line filter for removing spikes in the power line/220 VAC line on your SS Amplifier.


Gito

[W1RKW]

T,
Any DC on the output of the SS amplifier?

[N3DRB The Derb]

Tom,

how about a good old fashioned speech amp just up scaled a bit? If'n it were mine, I'd be protecting that iron at all costs, S.S. amp b damned. Aint gonna fret over no 2 bit amp - just zag instead of zig. 

[w1vtp]

Hi
 
<snip>
unkeying such big transmitter can cause a spike in the power line/220 VAC line.
This spike got into the SS power supply trough the AC line,maybe small but amplified by the SS Amplifier and drove the Modulator,and since its not loaded( the RF amplifier is of) than a big surge current gets into  the modulator,makes a spark jump across the Gap.
Try to use a line filter for removing spikes in the power line/220 VAC line on your SS Amplifier.


Gito

Italics and bolded text my emphasis

I think this is worth investigating. Is there some sort of line filtering that can be inserted for the SS amp AC  connection?  Another possibility would be to use another tube amp to drive the 8 ohm input to the backwards connected AF xfmr.  I'm interested in this problem because I'm considering using the same approach for my 813 x 2 - 813's xmtr.  I'm already convinced that I'm going to have to spend a lot of time figuring out the sequencing.

I have a similar rack except the superstructure is made of extruded aluminum.  I still like your plexiglass approach but think that some sort of shielding will be ultimately needed to keep RF out of the shack.  I think I will be putting the 2 x 813's mod tubes on their own shelve.

Al

[WD5JKO]

I can eliminate the HV 750 ms delay, cuz it's on all the time - I just key a relay to direct the HV to the particular rig I'm using - quick key up.

   Tom,

    I think the quick unkeying of the HV B+ is the root of the problem. You want that big 70H choke to decay at a L/R time constant. I'm having trouble visualizing your circuit, but If the B+ line floats after unkeying, when energy stored is yet to be released, then all sorts of stuff could happen.

    Some ideas come to mind such as a HV commutation diode placed where it will work to aid the L/R energy dump and yet be transparent during normal operation. I also might have another relay ground the B+ line through a 1K (large wattage) resistor whenever the upstream HV relay opens up.

   Also keying HV DC is tricky business. Most relays will fail this test except for very exotic and very expensive ones. What kind of relay do you use to key the B+?

If you can control the transient, then maybe the AF SS driver amp won't be affected.  What comes first, the Chicken or the egg? 

Jim
WD5JKO