Bad Audio Driver Xfmr or...?

[Todd, KA1KAQ]

First Hosstraders, now this. I had this nagging feeling lately that the old rig (KW-1 for those not familiar) was a bit too trouble-free for having sat dormant so many years.

Was working a few AMers Saturday eve during the AWA event when it  appeared. I completed a contact with N9OO, took a break, came back to find the mod cathode current meter reading about 150ma on key up with no movement. Checked things over, shut down for the night, tackled it Sunday.

First, the circuit rundown:

Mic input to 12AX7 AF amp to [6AL5 clipper out of circuit, bypassed] 12AU7 Phase inverter to a pair of 6B4s. From 6B4G plates to T-404 audio driver xfmr, to pair of 810 grids, to mod iron.

Removed the audio deck from the transmitter (easy part) to find 2 weak 5V4 LV rectifiers and a bit of tar oozing out from the bottom edges of the sealed driver xfmr. Close inspection seemed to show dust on the tar, like it had been there a while. Checked both windings and found continuity end-to-end, CT to ends, and no continuity between primary and secondary. This was 'cold', of course. Not a lot of ooze, but enough to notice.

After cleaning and reworking a few things and reassembly, I get the following:

 - modulators static at 50ma to spec

 - audio drives normally (good meter deflection on voice peaks) for a very brief period, apparently until something heats up. Advancing mic gains yields no big difference afterward.

 - 6B4s get very hot, driver iron does also. IIRC, these tubes always ran somewhat hot. Now they're '6AQ5 hot', but on a larger envelope.

Tested the 6B4s again, no shorts and they seem to be fine. Swapped out the modulators, no difference. Driver bias was in full CCW, or off. Not sure how to set this with the deck installed, manual only states that it should be set for -60v on the grids of the 6B4Gs.

I confess - it's been YEARS since I got into this or any rig very deeply, beyond changing tubes, resistors, or caps. So, I'm a bit confused as to where else to look, or to (re)start looking, or what could cause what. For example, could a blown mod xfmr cause these symptoms or is it too far down the line for this description?

My gut tells me that, for whatever reason, the driver iron shorts when it gets hot, and stops it all right at that stage. Why it gets hot or how it got shorted - age, over voltage, etc I have no idea. I've seen a few of these rigs with that particular transformer replaced, so it would appear to be an issue. Just not sure why.

In the meantime, I'll start looking for a replacement. But I want to be sure that nothing caused this transformer to go bad (if it even is bad) before shoving another one in. Any ideas of what else to look for will be greatly appreciated. Maybe a day away from it will help these eyes to see better, or differently.

[w3jn]

Betchya anything the problem is associated with the 6B4s.  If there was no bi-ass on them they're gonna draw a ton of current thru the driver xformer.  Perhaps the enamel insulation on the wire has been damaged and it could be shorting in there somewhere when it warms up.  Or you could be losing bi-ass on the 6B4s somewhere.  Telling is that the bi-ass pot was all the way off... that could indicate there's an intermittent in it somewhere.

Good luck with this one, Todd!

[WA1GFZ]

Check the output caps off the 12au7 phase inverter. I bet one is leaking putting positive bias on a driver making them draw excessive current. The tube could be saturated so no outputsky.
You could monitor the driver grid voltage to verify the tube bias is changing.
The excessive current due to saturated tube will cause excessive transformer current and cooking tar.
The driver is DC coupled after the phase inverter so it is either a bias problem due to a leakage or a bias problem due to stage bias.
Or you have a bad driver pube.
the 5V4 could be crapped out due to excessive current which may have protected the transformer by crapping out.

[k4kyv]

If the bias on the 6B4G's checks normal, try testing the transformer with 110 volts a.c.  You don't even have to remove it from the transmitter to do this. 

Pull the 6B4G's, and apply 110 volts through an isolation transformer if you have one.  Otherwise unplug the transmitter and disconnect any ground connection, including antenna coax, T/R relay lines, etc - ANYTHING external to the transmitter that could have continuity to the ground.  If the transmitter cabinet is sitting directly on the floor, make sure there is no continuity to ground, or else insert pieces of insulating material under the cabinet.  The problem is that even partial continuity to ground could cause false readings, since one side of 110v is also grounded.  It could even damage some component in the transmitter, in the event there is a short circuit somewhere between driver plates and modulator grids.

Now apply 110 volts across the primary of the transformer, by applying it to the each of the plate pins in the 6B4G sockets.  Just shove a piece of zip cord with the insulation peeled back, far enough into the Pin-3 hole in each socket to make contact with the plate pin connector.  Double up about an inch of wire at the ends if necessary.  Test with an ohmmeter to make sure you have continuity.  Now connect a 75-watt lamp in series with each piece of zip cord, and connect the other ends of the lamp connections to a 110 v power cord.  The lamps will act as a current limiting resistor in each side of the line, in case something shorts, and will indicate a short circuit by lighting up very brightly.  Plug in the cord to 110v.  The lamps should have a barely perceptible glow, if any glow at all.  Now, using an a.c. voltmeter, check the voltage between the 810 grid caps.  If the driver transformer is typical, about 70-90 volts should appear from grid to grid.  Now, using a clip lead, briefly short the 810 grids together.  The lamps should brighten up.  Don't leave the grids shorted for very long, to avoid the possibility of overheating the transformer windings.

Now measure the voltage from ground to each side of the primary, at the 6B4G tube socket plate pins.  Each side should read exactly the same, and just slightly less than half the line voltage (you may encounter a slight voltage drop in the lamps).  You can measure the plate-to-plate voltage across the primary by measuring the voltage between the pin # 3 in each socket, or between the tube socket sides of the lamps. Check the voltage at each 810 grid.  Each grid reading to ground should read exactly half the voltage measured between the grids.  If these readings all check normally, as above, leave the whole thing connected for a couple of hours to make sure the readings don't abruptly change.

If the transformer has an internal short in one of the windings, the lamps in series with the primary will glow brightly, and the voltage across the 6B4 plate pins won't measure close to the same as line voltage, and the voltage to ground may be unbalanced.  If the primary winding is open, the proper voltage won't appear at the 810 grids, or there may be no voltage at all.  The voltage from ground to each pin-3 of the 6B4's may not be balanced.

If the transformer checks out OK, look for something open in the grid bias supply for the 6B4's the 810's or driver plate supply.

I wouldn't worry too  much about  the old tar oozing out of the transformer.  That is not uncommon with potted transformers.  It is usually of no consequence.  If the tar looks fresh, and the leak appears to have just started occurring, that would be more a cause  for worry.

If it looks like it may be the driver transformer, don't throw away the old one until you have checked the transmitter thoroughly with the new one.  It could be something else, and those transformers are  too scarce to risk discarding a good one.  Better still, borrow a known good driver transformer, even one that is not exactly the same as the original, and JS wire it in.  That  should at least confirm that the original driver xfmr needs replacing.

Just any old transformer won't do as a permanent replacement.  The class B driver transformer is a critical link in the audio chain, and must have extremely low leakage reactance and internal resistance to maintain a "stiff" source of audio voltage for the 810 grids.  Otherwise, severe waveform distortion and splatter will occur.

If all voltages associated with the driver stage (filament, grid bias and plate), check normal, I don't see how a problem with the mod transformer could cause the driver stage to heat up.

Intermittents suck.  It can be a real pain to chase down the problem, especially if it occurs infrequently, at unexpected moments, for short durations only.  Yet that probably describes over 95% of the malfunctions I have ever had in my station equipment.

[Todd, KA1KAQ]

Wow, a lot of good input! Thanks guys.

Here's what I know to be true so far: it sure isn't an intermittent problem at this point, so at least I should be able to track it down. I don't remember there being tar oozing out in the past, but it's also been a long time since I was inside this thing. There wasn't a lot, but enough to peel off a small chunk which had hardened.

I was really puzzled by the driver bias being turned completely CCW also. Apparently it's been that way since I got it, unless Bruce cranked it around when he was helping me get it straightened out back in April. I don't think so, just wanna see if he's paying attention.

But the biggest problem is how to get under the deck to check the bias readings, muchless to set them correctly. The only way I can think to do so is removing the deck from the cabinet and setting it up on a table behind with mucho clipleads ala 'QIX. Other than the 230v line, nothing else will move or flex that far, and there's no room to get underneath with it installed. It can't be as simple as pulling a tube and checking the pins with a pointy lead, right?

So you've confirmed what I was feeling, sortakinda: hot tubes and hot xfmr probably related to problem in the 6B4 stage, possibly between the inverter and that stage (although I'm pretty sure those caps were replaced with SBE OD's in '92). I went through, over, and around the schematic so many times over the weekend I was seeing it in my sleep!

I'll look things over again tonight and probably remove the audio deck again. It comes out easily, going back in requires a bit more time for alignment.

[WA1GFZ]

I would use an isolation transformer and variac to apply an ac voltage from a main into a little driver transformer. I wouuld also use a small fuse on the line to protect the transformer.
you should be able to monitor the 810 grids with a scope to see if one phase drops out due to an internal short.
Changing the configuration just adds more variables. I would save Don's test until you suspect the transformer after you verify the driver stage is functional.  

[WA1GFZ]

Todd,
Tack solder a couple leads on the tube socket pins and bring them out to a test strip. Then you can monitor things going on. Clip leads can fall off and cause damage .
just because caps were replaced doesn't mean they can't fail again.
Say the cap doesn't have the right voltage rating.
A shorted driver transformer will not cause the drivers to run hot.
The DC resistance is low to begin with.
I would start with the driver grid voltage myself.

[Todd, KA1KAQ]

Sounds like a plan, Frank. Hadn't thought of tacking on leads. Makes good sense. I do know there are still a few (2-3) old wax and paper caps under that that need replacing, I ran out when I was reworking the audio back in '92 and then had a 4-250 crap out so didn't get back to it until a couple years ago.

If anyone is curious to see the schematic and overall layout, the CCA has a pdf of the manual online at:

http://www.collinsradio.org/html/manuals.html

[WA1GFZ]

7-4 Audio section C412 and C413 would be my first choice. Your Modulator bias current may go up slightly as they fail.
The Bias pots are in series so there will be a little interaction.

[WA1GFZ]

A shorted C417 would jack mod resting current and effect driver bias.
This is at the CT of the driver transformer secondary where bias is fed to the modualtors. Since the bias pots are in series a short would yank the driver bias voltage. 10 uf cap.

[W1RKW]

Todd,
Although I have no experience with this xmitter, I have to agree with Frank on the checking caps.  My experience with troubleshooting electronics over the years especially with stuff that is older and caps that have years on them typically I would immediately suspect an electrolytic.

Without getting into an OBT my experience with electrolytics is they will become leaky and I automatically suspect them when troubleshooting audio circuits.

On a side note, I bought a dozen 47uf 50vdc electrolytic's recently to repair a couple of switching power supplies. Half of them were out of spec and so I returned them for replacement. 

[Todd, KA1KAQ]

A shorted C417 would jack mod resting current and effect driver bias.
This is at the CT of the driver transformer secondary where bias is fed to the modualtors. Since the bias pots are in series a short would yank the driver bias voltage. 10 uf cap.

I bet you're right on the money, Frank. I completely missed this cap and another below it, both clipped to the side panel underneath. Also managed to miss (or not factor in) the possible impact running through the schematic. So hopefully it will be as simple as that, although I am going to replace the remaining caps I didn't change before, including all of the electrolytics.

Any thoughts on what changing the value of that cap might due to the bias? Say - from 10 to 20 uf? I'll go with a beefier voltage rating regardless, hopefully I can find one fat enough to fit the clip.

Bob, I'd be curious about what brand of caps you had trouble with. In fact, that might be a good topic for a new post instead of burying it in this thread.

Thanks again, guys. I'll report my results once I get the time and components.

[WA1GFZ]

Todd,
The bias is not regulated so be careful increasing the value of the 10 uF. This could charge to a higher voltage pushing the modulators closer to cutoff. A Bias regualtor is usually a shunt bleeding current off. The only to bleed current off the KW1 circuit is back through the pots. Increasing the value to 20 uF would double the time constant of the circuit so be careful. This could actually cause distortion.
Good luck fc

[k4kyv]

I used to have electronically regulated grid bias on the 810 modulator grids in my HF-300  rig.  It was very stiff - I could hit the audio with all my might, so that the waveform was coming out a square wave, and the bias voltage would change maybe half the width of the meter pointer.  The circuit is described in the 1957 ARRL Handbook and uses a 6SJ7 DC amplifier tube, a couple of VR-150's and a 6AS7G regulator tube.

But the problem was that the modulator plate resting current was very unstable.  I finally figured out what was happening.  With a.c. line voltage variations, the modulator plate voltage was varying.  But with the fixed regulated grid bias, the bias voltage was not shifting proportionally with the plate voltage.  The effect of this was a change in grid voltage relative to the plate voltage, and the modulator tubes, being amplifying devices, showed a noticeable change in resting plate current.

I took out the regulated bias supply, and substituted a "brute force" bias supply, running at 200 volts, 500 ma DC full wave rectified, into a low-resistance bleeder network.  It uses separate 500-ohm rheostats as part of the bleeder network to pick off bias voltage for each modulator tube.

The bias voltage regulation with normal audio modulation is almost, but not quite, as good as it was with the regulated bias supply, but now there is much less variation in resting plate current with changes in a.c. line voltage. If the mains voltage drops 10%, so does the bias voltage, which counteracts the drop in plate voltage, resulting in only a 10% drop in resting plate current.  With the old regulated bias supply, with a 10% drop in a.c. mains voltage, the resting plate current dropped more like 35-50%.

I suspect this is the reason most of the older plate modulated AM broadcast transmitters with the classic pushpull plate to pushpull grids class-B driver transformer, used a similar brute force bias supply for the modulator tubes and not electronic voltage regulation.

[WA1GFZ]

My point exactly Don. Better to not mess with the circuit or you could end up with some weird effects. The load on the bias supply would be modified with a larger hold up cap. The load resistor and two bias pots
(driver and modulator) are all in series in the KW1 circuit

[Todd, KA1KAQ]

Cool stuff, Don. I understood most of what you said, so either I'm remembering or re-learning. 

Update: Just ordered a whopping $90 worth of new caps from AES to replace the remaining electrolytics and wax/paper tubulars I didn't change in '92. The only values I stiffened were the multi-section cans used for plate decoupling (C404-ABC doubled from 20 to 40 mf) and a dual filter can (C415-AB from 40 to 100 mf). I did increase the voltage ratings on all.

While I still prefer the old 'if it ain't broke' approach, in this case I'm going to do them all since there are so few and there is so much iron on that deck to potentially fry. Skip also sent out a good replacement tranny via UPS 2 Day, so if all goes well and there are no major events in the universe, I could very well have this piggy back on line sometime Saturday night or Sunday. I figure if 'JN can tackle that TMC brute which is FAR more difficult to work on, I have no excuse not to do the same on this piece-o-cake.

Also got an email from Tom, W2ILA informing me that he was sick of waiting for me to get the 300G back on 160 so he sent out a replacement tank coil which will allow me to reassemble it and hopefully make some RF with it before the move.

Stay tuned, results to eventually follow.

[AB2EZ]

Todd

As others have stated, in the KW-1, the negative bias for the 6B4G tubes and for the 810 modulator tubes comes from a resistive voltage divider hanging off of the -150 volt supply. If anything causes this bias voltage to drop (e.g., a low impedance to ground across the output of the -150 volt supply... then the plate current will rise dramatically on both the 6B4G's (drawing excessive current through the primary of the driver transformer) and the 810's (drawing excessive current though the primary of the modulation transformer).

Here is a suggestion for how to track this down:

Remove both 6B4Gs (no tubes, no 6B4G plate current). Turn on the filament switch, but do not turn the plate switch on (no plate voltage on the 810s, no 810 plate current).


Measure the grid bias voltage on one of the 810s (between the grid cap on one of the 810's and ground).

It should be around -60 volts...depending upon various factors such as the types of 810's you are using and the normal plate voltage that would be applied to the 810's during regular operation. Remember, the plate switch is off, so there should be no voltage on the plates of the 810s.

If the bias voltage on the grids of the 810's disappears... then look for the obvious places that could be loading down the the bias supply. Also look to see if there is a cold solder joint somewhere in the -150 volt supply.



Best regards
Stu

[WA1GFZ]

Stu has a good point. If the voltage changes also check the 10 UF cap at the center tap of the driver transformer secondary. Then you can also monitor the driver grid voltage to see if it changes. A leaky cap off the phase inverter would swing the voltage due the high series resistance back to the bias source. A change in driver bias would only have a small effect on modulator bias voltage for the same reason. The driver tubes removed would help to isolate the problem without causing additional damage

[Todd, KA1KAQ]

Good idea, Stu. I'll give this a quick peek before I tear into it. The caps and tranny both shipped yesterday, it's just a question of whether they'll get here by the weekend. Judging by recent experiences with USPS Priority (formerly 2-3 days, now 4-8 or...?), probably not.

[flintstone mop]

There was a very good circuit in ER mag to eliminate the driver transformer misery. You can still feed biass to the pubes and there's only about $20 worth of resistors and condensers, if this is a push-pull audio outpoot your are driving.
Fred

[WA1GFZ]

How dare you suggest molesting a KW1 Fred.

[flintstone mop]

But, But
It would be under the chassis. No one would ever know. Just keeding
The leaky caps from the driver tubes is always a good starting point.
Troubleshooting is good excersize for the brain cells
Fred

[N3DRB The Derb]

if it's under the chassis and the stock components are left in place it's no foul. Start drilling and blasting holes and stuff and yeah, that's gay, but a sub like that
that doesn't show and can be reversed at any time is fair game even on a KW-1. I subscribe to the slab bacon rule in that regard. If you cant see it above the deck (no new holes, no new components, no cosmetic changes) then it's fair game as long as it's reversible.

there are people who scrape the filling out of the old wax caps out and put new caps inside the 1930's casing on radios. So they look exactly like they did - I'm not sure why they feel the need, but lots of them do it.

These are radios not religious shrines. Art Collins wont get mad and haunt yer ass if you drop the driver transformer.

I'm just sayin.

[w3jn]

These are radios not religious shrines. Art Collins wont get mad and haunt yer ass if you drop the driver transformer.

Blasphemy!  You should have seen the hue and cry when I showed how to fix the endpoint adjust on a 51J PTO by removing a turn...

[WA1GFZ]

OK OK I'm just trying to crank you guys up.
I modify everything I own.