The AM Forum

I would like to convert my Speach amp from AC to DC filiments. I have found several circiuts on the web.


One way is to use a simple hum balance pot.    Term 1 goes to one side of fil trans secondary. Term 2 goes to ground and terminal 3 goes to the other side of the fil trans secondary.  A 100 ohm pot is suggested.  You can dial the hum away.



The other system is to use DC offset.  You make a simple voltage divider off the B+ to get around 4 volts DC to the center terminal of the 100 ohm pot.  Then you can adjust the hum to zero.

I guess also you could use a Diode and a cap to make a DC supply for the filiments but this is not suggested in my cases. I guess that is because the voltage will be changed.

What circiuts have you used?  Any luck?

Neither one of those adaptations is a DC filament operation.

They work fine, however, as an out-of-phase additive to get the hum nulled out.  This method is a lot easier to do than working up a filtered low-impedance DC source.

Cathode tubes are easier to run on DC than filaments, as there will be no tendency to pull more electrons from one side of the filament, causing a shortened life.

Lots of audio equipment uses out-of-phase global feedback to reduce noise as well as hum artifacts.

Vintage radios that used field coil speakers sometimes had a secondary winding there to lower the hum level right at the speaker.

73DG

Yes.. Of course a 100 Ohm pot is not producing DC.   ::)

So your suggesting the hum balance pot and or some feedback?

C

Clark,

  It would be helpful to know the tube types used in the low level stages of the T368. Also are the filaments grounded to chassis on one side? Is the chassis steel or aluminum? Are there any audio transformers in the low level audio chain that are near a power transformer or a filter choke? If so are these transformers encased in iron , aluminum, or open construction.

  I am asking because the solution to your issue depends a lot on your unique situation.

Oh, is the input audio return tied directly to the chassis, or is it tied to some sort of audio common that might have an impedance (like 10 ohms) to chassis ground?

Jim
WD5JKO

I do recommend the 'electronic' method of limiting hum rather than the brute force track.  

A simple thing to try is flipping the polarity of the primary on the power transformer feeding that stage.  It may very well be causing an 'additive' hum, and trying that may get the 'subtractive' reduction you are looking for.

73DG

Dennis. I will perform this test today.  Thanks for the tip! 

Here are the screen shots of the 60, 120 ect on the SDR radio receiver.  Both transmitters where tested into the same dummy load. The SDR was hooked to the antenna switch nearby. 


Keep in mind that I am zoomed in to the lower portion of the audio spectrum and that I have used the peak slider to make these visable as large peaks.  Use the DB scale at the right to compare noise floor to PIP at 60 and 120.

Icom 756 Pro has 20 DB or so on the 60 cycle and 35 to 40 on the 120.
The T368 has about 55 to 60DB down for 60 Cycle and 55 to 60DB on 120.

Nobody has ever mentioned hum on the Icom ever.   Can someone please let me know:

1. Did I read the plots correctly?
2. Are these figures well below what is considered a clean signal?
3. At what level can signals like this be heard by the casual listener?
4. Am I waisting my time trying to further reduce the 60 and 120 on the T368?
5 Should I tell the freaks that send flex radio plots to my inbox of my transmitters they should jump off a bridge?

Jim.  The Preamp tubes are 12au7s  Yes they are by sealed potted Crosley/saratoga transformers. I think Copper or aluminum cases.  I will look over the schematic again for the wiring question. The T3 preamp is known for all kinds of hum.

If I pull the two preamp tubes, the Hum goes down to a point where I cant even hear it here.  But there is still a blip on the plot.

Anything more than 40 dB down isn't worth worrying about unless you just want to see what can be done.

Some cathode type tubes hum because of a high-resistance path, heater to cathode leakage. It's obviously less annoying in their level stages than lower.

Heater to cathode leakage due to the inherent diode, and or hum from the heaters, can sometimes be reduced by applying a DC bias to the heater circuit so it is 25-50V above ground. Hi-fi amps frequently do this. Some take the voltage conveniently from the top of the output stage's cathode bias resistor.

Some audio guys run directly heated tubes from DC rather than the old way of using a transformer center tap for the GND.

About the DC polarity on a filament eating one side of the filament coating faster, isn't one solution to periodically reverse the polarity? Just have a latching relay that swaps it each time the gear is turned on.

Some modernistic folks will run their directly heated tubes from a high frequency AC source, 100KHz or some frequency that is not audible. That takes yet another special power supply.

Another thing I forgot to mention is that grounding the CT of the filament supply essentially makes the hum frequency half, or 30Hz on a 60 Hz supply reference to ground.

That low a freq is easily not a player in most uses, as audibly reproducing it is not easy.

73DG

I run the filaments on my old mic preamp, and the VFO tube, off the same DC supply.  I once had a neat little hermetically sealed regulated DC supply that put out 6.3 volts at about 3 amps, but it crapped out.  I replaced it with the 12 VDC supply that used to run my 2m CB rig (which I haven't used in years), with a resistor in series to drop the voltage down to 6.3.

The VFO filament was  more for the regulation than for the DC, since even a volt or so change in  line voltage caused my VFO to shift enough to move the frequency 100~ or more on 40m, which was a problem when I tried to work 40m CW. The single-ended 12AX7 mic preamp had loads of hum with a.c. on the filament, but it went away when I ran it off DC.

The push-pull pre-amp and balanced output D-104 combination I have been using lately works just fine running the preamp tubes on a.c.

I run my 12AY7 mic amp filament off a 21 volt DC supply.  Voltage on the filament runs from about 21V down to about 15V.  The remaining 15 volts is dropped through a resistor to ground.  So the filament is about 15-20 volts DC above ground which also helps to reduce hum. The same 21v supply is used for the 12VDC relays using dropping resistors.

The filaments for the VFO tubes run off a regulated 6.2 volt DC supply.  This eliminated the minor drift due to slight line voltage changes.

Fred

Clark, any chance you can post the T-368 speech amp schematic?

Here is a great quote from Bruce W2XR from 2008 (HumApache thread):

"And do all of the audio grounds return to a single common point very close to the audio input connector? I'm just wondering about the possibility of a ground loop here. A 60 hz hum can frequently spell a ground loop. When looking at the hum pattern on a scope, if it is not a clean 60 hz sine wave, this is a good indication of a ground loop. If you have multiple grounds within the audio section of this rig, you are asking for trouble. All of the grounds must return to a single common ground point, using the popular star-ground topology; you can't take the expedient path of using the grounding lugs on a tube socket to ground the components for that particular stage."

Many good points brought up on this thread including how DC filaments don't always fix the problem; topic ended though without resolution:
http://amfone.net/Amforum/index.php?topic=13900.0

Looking back, I bet the hum was magnetically coupled to the open frame audio interstage transformer, that is if there is one on the Apache..

I like the tube substitution of a 5814 (Low noise and microphonic rated 12AU7). Might help depending on the cause of the hum.

Once in my QRO Central Electronics 20A I had magnetically coupled hum from an added filter choke to one of the audio interstage transformers. I experimented with a hum neutralizing circuit consisting of a resistor (about 22 ohms 2W) off one side of the 6.3 VCT filament winding to chassis ground. The magnitude of the induced ground current, phase, and location to chassis were all important. I could make the hum far worse, and considerably better with experimentation. If the hum had been pure 60hz and sinusoidal, the technique would have worked with a deeper null. Instead I had to replace that filter choke with another that was encased in a steel can. Hum gone!  



Jim
WD5JKO

How does that work? It's push pull -so 120Hz seems more likely

The famous Collins PTO and its clones drift substantially with even a minuscule change in filament voltage.  The regulated DC on the osc. tube cured my VFO (a highly modified T-368 master oscillator/multiplier unit). I had the exact same problem with the 75A-4.  I use an updated version of the Sola constant voltage transformer to run the whole receiver.  It is relatively small, about 3 X 4 X  5 inches, the buzz is just barely perceptible and it runs cool. It was designed to run a PC, according to the vendor who sold it to me.

One advantage of a FET or other solid state vfo.  No tube filament voltage to regulate or filament hum to suppress.

I got an old Sola from my employer on eBay.  That thing made so much noise and threw off so much heat I hated using it and eventually gave it away.

I ran some more tests this morning.  Hooking up the Green wire on the power to the rig lowered the 60 hum by 8 DB.  I have the green wire from the wall socket and I have an seperate station ground system.  Hooking up the station ground does nothing. But when I hooked up the green wire, The hum did Drop a bit. I guess I will leave that on.

I think I might try swaping the HOT and Neutral line on the plug next. 

The remaining hum can be varied by changing the modulator BIAS control on the front panel.  If I bias the tubes off, The hum is gone.  If I turn the current up to 150 to 200 MA, Then it sure does get noticable.   

I think I will have to go into the preamp to get anything lower now.  I am considering bypassing the first stage.  Running TWO 12au7s like that back to back is a tremdous amount of gain.  I think its simply to much for anyones sake. 

Keep in mind that the rig came with a 12aT7 which then fed a 12au7.   If I run this way, the hum is super strong.  Removing the 12AT7 and inserting a 12au7 reduced the hum way way down. 

I will get a schematic up here shortly!

C

Here are the schematics Patrick. 

I have the "A" model Speach AMP.  This was a one time run.  The reverted back to the basic design and used that through until the F which was changed again I understand.

I have included TWO  schematics.  One of A and one of the other design.

If you notice that on the "other" SA's they used two 2.5 chokes on the input.  My Speach amp does not have those chokes.  Only resistors.

Also.  On the Filiment supply,  There are two .01 caps but only on the First and second preamp tube.

C

Some of the AC regulators are noisy and they are all inefficient when not under a decent load. When the load is near full, the buzz is less and the waveform is nice (nicer..) The advantage is low cost and simplicity.

Compare the cost of a Sola mains regulator to one of those variac-controlled buck boost units, it's 5 to 1 or more. Comparing it to a GE Inductrol the difference is even more extreme.

here, there are one or two Sola and similar-type filament transformers on the shelves, as well as a few little 100-200VA ones some OM has put on a chassis with a plug and socket. I am saving them for projects. They are small and I don't think the annoyances will be noticed. Those kind of small Sola regulators are pretty innocuous.

My low voltage rack has two 24V/25A Sola DC supplies in parallel for working on military gear, and a 48V/5A one. I am very happy with them. I do not mind the buzz or the inefficiency but they don't run all the time. $20 each - It was far cheaper than buying active-device regulated supplies.

I'm just saying don't knock them all, and understand they have to be well loaded to avoid the annoyances. One thing they do well is eat large AC voltage excursions that would otherwise blow up the served equipment.

I don't see anything there that ought to make a difference. The RF chokes should not contribute to hum unless they are ina  field and have a low-Z connection at the input end.

But you mentioned 4V as a bias on the filaments. That is interesting because I have always been told at least 25V and as much as 50V. If the hum is in any way from a H-K emission or the like, 4V is not enough to resist 6.3V.

Have you tried operating the unit, and then removing and restoring heater power from the suspected tube? An A-B test, As in, setting it up where you can disconnect it completely (both leads), even with a DPST switch. The hum should either do nothing or go away, or, anyway something has to happen.. It's an experiment, maybe a pain to do but it should give an answer as to whether the hum is due to filament power or something else, on a per-tube basis.

At one time I had a couple of those "test extension sockets", a 7 pin and a 9 pin, that I'd opened and wired with a DPST switch for exactly that purpose, because customers would complain of hum in their amplifiers and about half the time it was the tube. But a tube checker would not catch it.

The pic is of a MX-1258/U extension set.

I looked at those schematics you posted.  Not sure what they are but, I would get rid of those filters with the iron core chokes.  I once built a filter in one of my HB rigs.  It used a iron core choke.  It picked up a lot of hum from other xfmrs on the chassis.  I tried everything to get rid of the hum pick up.  Finally I had to remove the choke and filter.

You also mention something about the bias voltage affecting the hum level.  Make sure your bias supply is well filtered.

Fred

Like Jim said, is it magnetic coupling? ? ? ? Some years back I had a situation that drove me absolutely NUTZ! ! ! ! ! I was using my DX-60, and then my little 5w piss weaker to excite my Junkston T-bolt. I had a terrible hum problem that was driving me crazy. First I tried DCing the filaments in the VFO - no help. Then I DCed the fils in the DX-60 - still no help. Then I chased it down looking for anything that might be causing a ground loop - STILL NO HELP!

I was about to reach for the BFH when the revelation came. I had been seeing the hum on the scope, but I hadn't actually listened to myself, so I put on the 'phones and starting listening to myself. Then the revelation came.........

I forgot to hit the plate switch on the T-bolt and there was no hum! ! !  As soon as I hit the plate switch the hum was there. Hmmmm............ Then I noticed that if I moved my chair, the hum would change  ???  ???

It ended up being the Dynamic mic that I was using at the time was picking up the magnetic field from the T-Bolt plate transformer! I had been had! ! ! ! !  

A little re-arranging of things totally eliminated the problem! And then I eliminated the T-bolt entirely just becasue it was a worthless POS......... ;D  ;D

I wonder if a T-3 could be picking up the plate transformer field with one of it's own interstage transformers, or something similar.

I have heard of using DC bias on the filament before, but never tried it. Of course, that would work only with indirectly heated cathode tubes.

Should the DC bias be positive, or negative, or does it make any difference?

For reducing hum I have had worthwhile results using an isolation transformer, so that both sides of the mains voltage are floating from ground and the equipment is isolated from the mains ground (green wire).  This may eliminate otherwise baffling ground loops.  Sometimes it helps to use an isolation transformer with a mid-tapped secondary with the tap grounded so that each side of the line is balanced, 60 volts above ground. Treat the 120v a.c. line like a balanced audio line. Try both the mains safety ground and a separate ground rod as your grounding point (for safety's sake do this only when using an isolation transformer), and see if it makes any difference.

If the iso transformer secondary is not mid-tapped, you can simulate one by bridging a mid-tapped choke across the output. The CT winding of a transformer, such as the HV secondary of a small power transformer or the primary of a push-pull audio output transformer, with all  the other windings left unused, will serve as an excellent choke for this purpose. Another possibility is a small variac.  Set the output voltage to mid-range, as measured with a voltmeter, and ground the variable tap. With the variac, you may be able to make a small adjustment to the mid-tap setting and null for minimum hum. Again, do this only in conjunction with an isolation transformer; otherwise you will probably blow a fuse or burn out the variac.

A DC bias on directly heated cathode tubes is done all of the time in grounded grid amplifiers. It actually works out about the same as applying the bias to the grid.  Also in an ART-13 (which uses DC on all of the fils) uses the connecting points of it's series-parallel connected filaments to pick off the bias for the 811 modders. And maybe even the 813 (dont remember for sure).

Applying a bias to indirectly heated tubes, in theory should make no difference as the filament is "supposed" to be insulated from the cathode.

but if there is some heater to cathode leakage (which isn't supposed to be there) it could well make a difference. Capacitive and maybe even inductive coupling of the fils to cathodes could, however induce a hum from the AC potential on the fils from small amounts of stray AC that end up "leaking" through and showing up on the cathodes of high gain stages.

1. So every one agrees the circiut with the 2.5 choke used on the previous and later speach amps are NOT needed?  I was thinking of adding them.

2. What about the 01 bypass caps on the Filiments at the first tube?

3.  Would you try the hum balance pot on this circiut?

4.  DZT,  I bypassed the High pass filter and the HUM is very loud. That filter has 50DB supression at 200hz and below. I think it said 60 or more at the 60/120hz point.  I cant bypass it. If i do, The hum is super super loud.  I do have the LOW pass filter bypassed.  It chops the Freq off at 2500 and up.  The goal is to reduce the hum and then bypass that filter.

One thing to NOTE is that printed ON THE FILTER is instructions to "Rotate" the transformer to reduce hum pickup!  Do they mean to flip the leads???  Or do they mean to ROTATE the Iron in the mount???  I found that to be odd!


Frank.  I have had the same kind of things going on here at time to time!   I found the problem was my Laptop power supply.  When I unplugged it, the station worked normaly again.  Lots of odd things can happen!

  They mean to rotate the iron core! Must be one of those transformers with a single centered mounting thread and a jam nut. If so, this might be the "Bingo" you were looking for. A Campbell's soup can (if magnetic) over it might also lessen the hum.

Jim
WD5JKO

<snip> "One thing to NOTE is that printed ON THE FILTER is instructions to "Rotate" the transformer to reduce hum pickup!  Do they mean to flip the leads???  Or do they mean to ROTATE the Iron in the mount???  I found that to be odd!" <snip>

Hmmmmm............ Sounds like they had that problem all along It must have some magnetic coupling to something else. Is there a fil transformer close by??

Many older designs used structured frequency response to eliminate the trash that made it's way into the audio. This was a very common thing that rears it's ugly head and bites you in the ass when you try to produce HI-FI audio. You end up opening up the floodgates to the garbage. Most of the all stock ones sound like the operator has a clothespin on their nose.

Unfortunately, I am not familiar with the innards of a T-3, but I have heard some that sound real good. I wonder if they have eliminated the hum by not using the low-level stages and feeding them with line level audio further up the line. If not, the soup can testing may be the next step. (Or find someone that has a good sounding one and ask them what they did)

I would get rid of any iron core chokes in the low level audio stages.  Just by-passing it while it's still connected in the circuit probably would have made the hum louder.  You need to cut the whole filter out and place it over on the table someplace.

Rotating means just that, rotating the core to reduce hum pick up.  Minimum pickup occurs when the fields are at right angles to each other, in theory.

Fred

Ok.


So Next step here is to get a Sat when I am bored, then modify the amp. I need to remove the Filters from the chassis, Braid the filiment wires and install Hum balance Pot.  Then back in and test again I guess.

I wish I had another speech amp to install and then I could just gut this one and install the Circiut out of the 1950s handbook and start Fresh.

C

While thumbing through the manual,  I noticed I overlooked TWO Can capacitors.  C17 and C18 on the modulator deck.  One is 1,100 UF and one is 400UF.  Voltage is not given. 

These caps are hooked to the center tap of the filiment transformer for the Bias supply.  It looks like it might be a source of bias for the carbon mic?

Can someone please help me and look at the following schematic for C17 and C18.  They are 60 years old, I am guessing they are shot. They are Typical twist lock malory can capacitors, I cant read the voltage rating on the caps.

I have caps here near those values in 63 volt rating.  Do the values need to be exact?  It looks like some kind of half wave supply.  If those are dead, Could this contribute to the hum? I cant imagine they are good after 60 years! 

C

C17 and C18 are very important. The current waveform through R19 would be a "fullwave rectified" sinewave, like raw DC, without them. If they are bad, the carbon microphone will receive its DC current with a generous helping of hum.

As for the bias for filaments, I have always read the filament should be positive to prevent leakage current to the cathode. It has something to do with the way the parasitic heater-cathode "diode" works. I'm not sure I ever saw a detailed scientific explanation until now, never looked for it just believed. It's an ancient practice. The link mentions the RCA RC-26. The attachment is the paper from 1936.
http://www.diyaudio.com/forums/tubes-valves/187134-elevated-heater-voltages-40v.html

As for those RF chokes, maybe they are only needed if there's a problem. If your set did not have them, why add them? As for their metal, a ferrite core choke might be better than an iron core one for rejecting hum but that is only an opinion. It's a question for a transformer expert, not me!

So C17 and C18 are only for bias supply for carbon mic?  The carbon mic works great with the same low level hum that the rest of the rig has.  Only, You sound like you are being called to the principals office  ;D


What voltage would you suggest as replacements?  Its confusing because it shows them hooked to the hv supply. They must be low like 50 or less.  They did not make HV caps in those values anywhere near that size in 1951.

As for the chokes..  The Basic and first version had the chokes.  Then, this A model removed the chokes.  The B and all other versions had them installed.  Its like they pulled them out and then put them right back on the next model.

I ordered some. I might put them in. Why not? 

I am very confident that the low level hum is the Filter/filiments in the 12au7. If I remove the 12au7 the hum is gone.

How about a I sacrafice an old 12au7 and snip its Fil pins OFF>  Then plug it back in with no fil power as a test.

C

The voltage rating of those two caps would be low.  Your 63 volt caps would be fine.  I think you should check the two 8ufd caps that are connected to the two 6hy choke.  They would be high voltage caps (450v).

I took another look at the schematic.  I don't like the fact that the two 8ufd caps are grounded back to the center tap on the filament xfmr.  Probably should have been grounded to the neg return or chassis ground if one is used.  Anyway, the two 8ufd caps are grounded through the high capacitance of the 1100ufd cap.  So if that cap is open you have no filtering on the mod bias supply.

I think you did mention that the bias voltage affected the hum.

Have to take another look, not sure about my comment above.

Took a third look.  The two 8ufd caps are grounded through the 100 ohm resistor.  That doesn't make much sense since there will always be some ripple left unfiltered.  The 8fd caps should be grounded at the ground point of the 100ohm resistor.

Fred

Fred, Thank you very much.  I had read that section of the schematic half a dozen times last week scratching my head.  WHy did they use a 100 ohm?  This makes no sense to me. 

1. I have some 1000UF at 63volt caps here. Will that be sufficient for the replacement of the 1100?  I can test them all and find the high one in the batch.

2.  I have some 250UF at 50 volts caps here.  I will use two for the replacement of the 400 uf can. I will find an equal match.

3.  I will move the ground point to the other side of the 100 ohm or chassis ground for both 8UF caps.  I have already replaced the caps with 22 UF 500 volt electroylitics.  This change is simple and will take moments.

Sound good?

Moving that ground point may or may not make any difference, give it a try.  The 1000ufd cap should be OK, not much difference between 1000ufd and 1100ufd. a few 250ufd will work fine for the 400ufd.  The voltage drop across the 100ohm resistor is for (I think) the carbon mike.  

That whole power supply is nothing more than an ordinary FW power supply.  The difference is that they use negative lead filtering.  Usually the ground point would be at the CT of the HV winding. This would make the whole supply voltage positive with respect to ground.  The most negative point is the HV winding CT.  The most positive point is the CT of the filament xfmr.  Even if there were no ground point you would still read the full output voltage of the supply with a volt meter placed from the HV CT (neg) and filament xfmr CT(pos) (400-500v).

There is a voltage divider across the whole supply (3500ohm, I think is was 350ohm and the 100ohm resistor).  Now add the ground point somewhere along the voltage dividers.  In this case it is at the junction of the 100ohm and 350ohm resistors.  So you can see that with respect to chassis ground the HV CT is still more negative than the chassis ground. Hence you have a negative bias supply with respect to chassis ground.  From the chassis ground point,  the filament xfmr CT is still more positive than chassis ground. Hence a positive voltage for the carbon mike bias.  A carbon mike needs some current flowing through it to work.  This positive low voltage also has its own CRC filter circuit (1100ufd, 100ohm, 400ufd.

Because the 1100ufd cap is across only a 100ohm resistor with nearly 4000ohms in the whole voltage divider, you can see the voltage division is about 40 to 1.  With a total voltage from the supply being a max of about 500 volts you can see that there is probably no more than about 15 volts across the 1100ufd cap.

Hope some of this helps you to understand your power supply circuit.

Fred

Interesting circuit........... the 2 8uF caps are the inpoot filterz for the whole shebang but..... The positive side of that supply goes to ground through a voltage divider circuit, to provide a low B+ for other things. The 1000 and 400 are fairly low voltage caps. If you are going to buy them, I'd go for 160v caps just to have a little cushion factor. If bad, they could cause a hum issue. Be sure to replace the 8uF caps as well as they are filters for the modulator biass.

I have seen similar circuits in old 30s broadcast receivers, Philco was famous for them, except they put it in the negative lead (seondary center tap) to ground side to obtain some biass for the outpoot tubes. Also check all of the associated resistors to make sure they haven't gone high.

Just a return note on this one. When I did the Invader 2000 that was one of the biggest problems I had to deal with. Once I opened it up for some bass response,
everything started showing up in the audio. The factory "eliminated" the problem by narrow banding the living hell out of the audio. It was quite a can of worms to deal with...........

Fred, Frank.
Thanks alot for helping me out!   I understand how this works now.  I was scratching my head on why they would have that 100 ohm resistor there for the ground.  I want to keep the carbon mic working and make this rig look stock.

I will yank the deck out again after work and install the 1000uf/63volts caps under the chassis and maybe I can find a high reading 250 to replace the 400.  Otherwise, I can series two 1000s and find a combo thats near 400.  I ordered some 100 UF 450 caps and received 1000s/63. The seller on ebay told me to keep the 1000s and sent me 450s.  I got 20 of them so why not use em!

I will also check those resistors frank.  Maybe one or more are shot.

The rig is working great now guys.  The hum level is low and the rig is running great. But as mentioned.. If you remove that audio filter and add in the higher value cathode bypass caps or more fidelity, You OPEN the flood gates for TRASH and the sharks smell blood and come swimming in for MHZ's away  :-[

All of this is just for fun and to keep me busy and out of trouble.

THanks again!

C

The problem with the circuit is the 8ufd caps are grounded in the wrong place.  They should be grounded at the chassis ground, the junction of the 100ohm and 350ohm resistor.  Once they're grounded there you can see that now the 1100ufd cap has the correct polarity(two filter caps in series) to complete the 8ufd cap filtering back to the PS positive which is the filament xfmr CT.  The way the circuit was built the positive side of the 8ufd caps is connected to the positive side of the 1100 cap.  The 8ufd filtering is completed to chassis ground through the 100ohm resistor.  This will always leave some level of ripple on the mod bias supply.

The voltage across the 1100ufd and 400ufd caps is low, probably not more than 15 volts.  You don't need 160volt caps.

How do you figger, Phred? ? ? ? The 8uF caps are across the total of the rectumfryer outpoot. They are connected from the filament of the rectumfryer to the center tap of the secondary winding. They are filtering the whole high voltage outpoot of the power supply. Just the positive end after the filters is held up slightly above ground (the 100 ohm resistor) to allow to tap off a small of + to operate other things. That whole voltage divider is also in effect the bleeder stack as well.

They used that same bleeder / voltage divider in old broadcast radios for many years to tap off some biass for the audio outpoot tubes. They just put it in the negative side instead of the positive side. The negative side was usually insulated from ground with a cardboard tube and a phenolic mounting wafer and tied to the center tap of the power transformer secondary. Which was then grounded through a resistor. The biass pick-off point was the junction of the secondary CT and negative side of the filter can.
I have been zapped many times when that resistor opened up with time. If you were to directly ground the 8uF caps it DEFINATELY create a hum. I have seen this with my own eyes when I misconnected a replacement cap in an old broadcast set.

Frankie,  I disagree.  The way the circuit is designed there will be some level of ripple that can be measured from the chassis ground to the CT of the filament xfmr.  The thing you're overlooking is that the bias for the mod works from chassis ground.

Build a power supply and put a resistor in series with the ground lead that runs to the neg terminal of the filter caps.  Let me know how much ripple you can measure at the neg terminal of the filter caps and ground.

Fred

Phred,
          It shouldn't matter. The filter is between the positive output lead and the negative output lead (2 chokes and 2 8uF caps) All of the filtering for the HV side is between + and - period! A voltage divider between + and - why would you want to put the filtering between the divider and ground? ? that would induce a hum in the outpoot. Take a very close look at the skizmatic that Clark posted. It is right by design. Exactly the way that I would do it if I needed to build it. That basic circuit but flipped over to the negative side has been around longer than either of us.

Take a real close look at the schematic and put your glasses on and scratch your head a few times. I think you'll see what is going on. I do value your opinion and respect your knowledge, but............... You are dead wrong here. It should make no difference.

The 400 and 1000uF caps are put there because that small voltage shows up to the low level AF stages where even a very small percentage of ripple applied to the grids would be highly amplified as it passes through the audio chain. Basically a "just for good luck" deal.

I'm going to run a few experiments later tonight to see what exactly is going on with that type of circuit.  I'll let you know what I find.

Fred

After work i pulled the deck out. Cap c17 was open.  C18 measured 500uf. Super leaky. I installed a new 1200uf and two a 500uf.  The ground runs from a very large resistor. Must be 100 watt. It measured ok. Going to fire it up in a few moments.

Frankie,

I did a quick experiment.  I have a power supply that has it negative and positive terminals isolated from chassis ground, it's a floating supply.  The supply is well filtered.  I measured the ripple with a DVM set for AC.  The supply had nearly no ripple with the meter leads connected to the neg and pos terminals.  Just what I would expect.

I then try another AC ripple measurement.  This time I connected one meter lead to chassis ground and the other to one of the PS terminals.  I was reading the full AC ripple with this measurement, just what I thought would happen.  So, even though the PS is well filtered, when measured between chassis ground and either neg or pos terminal the full AC ripple is there.  If I grounded the unused terminal to chassis ground the ripple completely disappeared.

So what have we learned.  A well filtered supply that is floating, will still have full AC ripple between chassis ground and the supply terminals.  When one of the supply terminals is grounded the ripple disappears.

In the xmtr supply that we have been commenting on, the PS is floating except for it is grounded through the 100ohm resistor.   So with a 100ohms in series with the pos point to the chassis ground you end up with some level of ripple as measure from chassis ground and the pos point (filament xfmr CT).  The amount of ripple is somewhere between no ripple (zero resistance to chassis ground) and full AC ripple (open to chassis ground).

This residual amount of ripple appears at the pos point and also at the neg point (bias for the mods).  So even though the supply is well filtered, the whole bias supply has a ripple when measured from chassis ground.

This can be eliminated by grounding the 8ufd caps to the chassis ground point instead of the CT of the filament xfmr.  With this, you then need the 1100ufd cap to completed the filter back to the fil xfmr CT.

This is my 3 cents on the subject.
Maybe we can get some more opinions.

Fred

Ok.  I got the deck back in.  The hum level is now so low I think it would never be noticed.  Its only 20DB over the noise floor on the scope.  I can hear a little hum using headphones and listening to the monitor rxer.  It might not be 60. It might be 120. 

The hum does not vary with MOD Bias now.

The hum does vary with the Mic gain a little. I think that is expected because of the filiment hum.

I wondering about grounding that center tap..  Should I try it and what circiut change needs to be made?

C

Just to answer on cap voltage - the current rating of the transformer there is 220mA, therefore the most current that could pass through the 100 Ohm resistor in the resistor stack R19 R22 R23 is 220mA, and that would make 22V. Since it is possibly a rectified sine and would have a peak value, multiply by 1.414, and the minimum cap volts might be 31VDC.

the resistor R19 should dissipate no more than 220mA * 31V, or 7W.

The 8uF caps passing through the 100 Ohm resistor to GND won't have significant ripple added to them from R19 if R19 is bypassed by the 1100uF cap.

You could move the bias supply GND to the fil. CT and move the carbon mike power to the top of R19, but there is no benefit I can see doing that. It could have a negative consequence of coupling the mike power to the bias current.


Snipping the heater pins from a tube won't allow an a/b test for H-K leakage because the heater won't be hot and the tube won't work.

The 100 ohm is not a 100 watt. I meant 10 watt. Its one of those green flat resistors that mounts with two studs.

I think I will move on to the filiments in the SA.  All the tubes are grounded on one side of the tube socket and then one wire snakes around inside the chassis.  Kinda like a johnson rig.

I am considering making some twisted pair wires and run them to the fils for each of those preamp tubes.  I almost bet that solves the SA hum and will allow me to get rid of that Hi pass filter.  This thing is pretty tiny sounding with nothing below 300hz.

If that does not work. I am going to try the 100 ohm hum balance pot.
C

The only thing I would try is moving the pos wire of the two 8ufd caps from the fil xfmr CT to the ground point of the 100ohm resistor.  The reason I suggest this if you look at the polarity of the 1100ufd cap you'll see that the pos side is also connected to xfmr CT.  You have two caps with their pos sides connected together.

If the 8ufd cap's pos side was at the ground point it will be connected to the neg side of the 1100ufd cap.  This is the correct way to have them connected.  You end up with complete filtering of the bias supply with respect to chassis ground.  This is what you want.  It eliminates any ripple issues with the 100ohm resistor.  The 1100ufd cap completes the filtering back to the fil xfmr CT. 

This is the same thing as when you stack filter caps in a HV supply.  The 8ufd caps in series with the 1100ufd cap has no reduction in the amount of capacitance.

Fred

Clark,
        I wouldn't ground the center tap of the rectifier fil winding. You then wouldn't have a place to pick off the mic bias (and whatever else may get it's power from there). Why re-invent the wheel if you dont have to.

Phred,
         "in theory" if the power supply is "floating" there should be NO potential, AC or DC between either leg and chassis ground, short, sweet and simple.
But............ You always seem to get a small amount of capacitive coupling between the primary and secondary windings. Henceforth the 1000uF cap is basically a short at AC against the small amount of capacitance between windings.

Going back to old broadcast receivers that use that circuit in the negative side, they usually have one filter cap between B+ and the "floating" B- lead, then another one across the whole shebang. The one across the rectifier outpoot is the basic ripple filter, and the one across the whole shebang to ground acts both as another filter, and bypasses out any AC riding on the outpoot from capacitive coupling in the transformer.

I would also think that a mil-spec transformer like you have on a T-3 would have that calculated in and held to a close tollerance. (Spec writers are always looking for something to make themselves look necessary) Sounds like now, all Clark has to do is adjust the rotation of that one interstage core and he is good to go. 

I have rotated power transformers, and "played  chess" with them to find the optimum spot on the chassis, and optimum orientation, to produce minimum hum pickup. The same thing works with the resonant coils in a multi-stage transmitter. Once the coil in the first or last stage is located, the coil in the following or previous stage is temporarily resonated with small variable cap and moved around and rotated until an rf sensor indicates minimum coupling.  Then the first two stages are fired up and the next coil down the line is treated the same way.  With multiple stages and multiple tuned circuits, each subsequent coil may have stray coupling to more than one previous coil, so the location and orientation may not be where you would think.

This technique is especially useful in open breadboard construction where each stage is unshielded from the rest.

With an audio amplifier, a transformer may be picking up hum from multiple sources as well, even from a nearby piece of equipment.  I once had that problem when my speech amp was located on the table adjacent to the receiver.  Moving it some distance away significantly reduced the hum.

Frankie,

All he needs to do is move the pos side of the 8ufd caps from the fil xfmr CT to the ground point at the 100ohm resistor and it will solve any issues with ripple across the 100ohm resistor.

I'm very familiar with the voltage dividers used in old BC receivers and the way negative bias was developed.

Fred

Nada............

It will effectively then put the 8uF cap in series with the 1000uF cap thus slightly lowering the value of the 8uf filters. The 1000uf cap should be more than sufficient to remove any ripple on the + side output.

Someone else chime in, please. I'm getting tired of typing............... ::)  ::)

Frankie,

You're wrong.  8ufd in series with 1000ufd isn't going to change anything.  What's more important is that the 8ufd ground return will not be going through the 100ohm resistor.  And the 8ufd caps will be on the correct polarity side of the 1000ufd cap.  Look at the schematic.

Probably will not make much difference but if I was building that supply I would ground the 8ufd caps at the chassis ground point.  The bias supply works from chassis ground and that's where the caps should be grounded.

Just because the xmtr was design that way doesn't always mean it's right.  Especially, a xmtr with so many hum problems.  That in itself should tell you something, like how many other errors are in the design.  The use of the high pass filter to reduce the hum problems the xmtr has is another indication of design issues.

Fred

He said after replacing the carbon mine caps that the hum level was now so low it would not be noticed. Isn't the problem solved?

Its a learning process.  Nothing is every fixed in a ham shack Pat. You should know that  ;D

C

Do you mean me, KD5OEI Patrick, or are you replying to N4LTA Pat? got to keep things straight around here. Or keep them confused. Or something.

I do know nothing is ever fixed in my shack. But then I like to stand in the middle and revel in the scattered parts.

Phred,
        This is my last post on this subject, done finito! ! ! ! You have A testa doro! !

1. that thing was designed by someone that knew more than either of us! And it was designed for a specific purpose. (which they did well) They were also designed to be able to be operated by some one with minimal knowledge of radio theory.

2. de-emphasizing the audio bandwidth was a very popular way to 1. eliminate noise monsters, and 2. "increase talk power" back in those days. It put the max volume right in the middle voice frequency range. (why waste mod power and final dissapation on frequencies that dont help your cause) and 3. minimise bandwidth.

3. the Hi-FI conversions are OUR thing, because we want to sound good and sound natural. Most, if not all military rigs with carbonium microphoniums are designed to sound like you have a clothespin on your nose.


Power supply filtering is always supposed to be on the rectifier output, period. Between + and - short, sweet, and simple. Holding the initial filtering above one side with a resistor is inviting a ripple issue. After the initial filtering you can do what you want with either side and voltage dividers, etc. But I have never seen any kind of power supply that had any resistance between one end of the initial filter and the rectifier outpoot.


Done, over, last post! I know you will want the last word!

 ::)  ::)

I meant you patrick.  ;)

C

Frankie,

Don't get your nuts in a uproar.  We'll have to just disagree on this issue.

Fred