DC Filament Supply Kicking My @$$

[dorrisant]

I have two Altec 1567As on the bench. I rebuilt both of them. One has been completely gutted and rewired, one has had all of the components changed. The original PTs were used.

At the owners request I installed a SMPS for the filaments. Not a good idea... Maybe because of the cheap crap he bought off of eBay. Terrible noise but he was willing to pay for the experiment.

Next we tried a few bridge rectifiers with a filter cap (lots of different values) then through a dropping resistor. The voltage worked fine at 12v but there was ripple on the scope and of course it sounded bad. I revisited this later and added different caps after the dropping resistor. Still looked bad on the scope.

I pulled that out and subbed in my Astron 13.6v 15A supply... Absolutely flat on the scope and no hum. OK. Now I know that it is the filament supply for sure and that it can work perfectly if that ripple is gone.

I told the customer that the original circuit as it is designed has been just fine for decades now... So I put that in subbing in 1N5402s in place of the selenium rectifiers. I have used these before in similar instances and they have always worked fine. This supplied the 12v but had 60Hz hum... About .35vac. No good. I also tried different cap values and caps at the output... All had about the same results.

Went back to a bridge rectifier with several incarnations of 7812 regulator circuits. Even nastier looking wave on the scope. At least the previous attempts produced near perfect sine waves.

I then tried a couple of different TXs just for the filaments so I could see if separating them from the PT would help... Again, no. Tried 6vdc too... Nopey.

I've been researching DC filament supplies for days and I'm really getting sick of it. I'm seeing 60Hz sine waves in my sleep. Seems that I'm just missing something, probably so obvious to others.

Now that I think about it the largest value of cap I used in any of these configurations was about 2000uF. Maybe I should have tried bigger values. I will have to try that.

HELP!!! Someone please show me the error of my ways! I want to sleep well again soon.

The no-load voltage on the original PT is 22vac. I also rebuilt a 1566A (only two 12AX7s as opposed to three DC supplied in the 1576As) just before this. That one turned out flawless! It's not a proximity issue I've moved the whole supply outside the chassis to prove it. One of the units has a bad winding, I believe. It shows relatively the same no-load voltage but drops off to a little less than 10vdc where the other unit will drop to 12vdc or just below.

If I had hair I would be pulling it out!

Tony

[tele_player]

Regarding the 7812 - you do realize that a 7812 is only guaranteed to provide 12V output when the input is 14.5V or above?

You mentioned that your rectified and filtered voltage was 12VDC with some ripple - a 7812 won't work under those conditions.

Robert

[martin manning]

I told the customer that the original circuit as it is designed has been just fine for decades now... So I put that in subbing in 1N5402s in place of the selenium rectifiers. I have used these before in similar instances and they have always worked fine. This supplied the 12v but had 60Hz hum... About .35vac. No good. I also tried different cap values and caps at the output... All had about the same results.

I'd expect 120Hz hum coming from a FW rectified supply so the 60Hz must be getting in somewhere else. Maybe there is a problem with the PT?

[Bob S]

Slightly off topic..
Don't feel so bad - I had a noise problem that I chased for 2 days.
Turned out to be elevated heater supply. Of course I changed everything else first despite the scope pointing me in the right direction all along...
You'll get it.

[Stevem]

A JJ ecc803 with its spiral filament should be all that's needed with a center taped heater winding so save yourself that pain in the ass and have that customer send that crap back for a refund if he or she can get it!

[martin manning]

FWIW, 22 VAC CT looks about right to deliver 12.6 VDC, and with the 1000u caps and 1R resistor you should see about 1V p-p ripple at 450mA (3x 12AX7). Bigger caps (4700u e.g.) will reduce the ripple proportionally.

A common 12.6 VAC filament transformer with a FWB would substitute nicely, and you could adjust the 1R to 2 or 3R to get the voltage dialed in.

[DonMoose]

I'd expect 120Hz hum coming from a FW rectified supply so the 60Hz must be getting in somewhere else. Maybe there is a problem with the PT?

Sounds like there's a bad ground somewhere - is there any ripple anywhere in the B+ rail after the first filter? How does Vscreen look?

[dorrisant]

Regarding the 7812 - you do realize that a 7812 is only guaranteed to provide 12V output when the input is 14.5V or above?

Unloaded I was seeing about 16vdc off of the rectifier. It held the voltage to 12.16vdc but with bad ripple.

Sounds like there's a bad ground somewhere - is there any ripple anywhere in the B+ rail after the first filter? How does Vscreen look?

Scope shows clean DC.

FWIW, 22 VAC CT looks about right to deliver 12.6 VDC, and with the 1000u caps and 1R resistor you should see about 1V p-p ripple at 450mA (3x 12AX7). Bigger caps (4700u e.g.) will reduce the ripple proportionally.

A common 12.6 VAC filament transformer with a FWB would substitute nicely, and you could adjust the 1R to 2 or 3R to get the voltage dialed in.

The bigger caps are the first thing I will try when I'm back at the bench and then the bridge rectifier if necessary.

Thank you all for the input. I feel that I was at least on the right path... Not just crazy. Thanks Bob, somehow sharing your painful story does help!

Tony

[LOUDthud]

Unloaded I was seeing about 16vdc off of the rectifier. It held the voltage to 12.16vdc but with bad ripple.

You have to look at the input with a scope when it's loaded. The minimum voltage at the bottom of the ripple must be greater than the 14.5V the regulator requires. Sometimes increasing the filter cap will fix this. Values up to 10,000uF are not out of the question.

Sounds like there's a bad ground somewhere - is there any ripple anywhere in the B+ rail after the first filter? How does Vscreen look?

There must be some kind of ground in the heater circuit. Otherwise the common mode 60Hz noise on the heater winding will not be suppressed. Connect one side of the 12VDC to ground.

When the heaters are cold they can draw two the three times the nominal current. Any regulator has to survive this without damage.

[gingertube]

And if you a re short a volt at the first filter cap then don't forget Shottky Diodes. These have lower forwrd voltage drop than silicion diodes and will give you maybe another volt in the raw supply.

I've had to use these before to fix similar problems of having just not enough voltage.

As an ASIDE on the subject of DC heater supplies.
Take a 5V AC winding intended for a tube rectifier heater and bridge rectify it with Schottky Diodes into 10,00uF/10V Cap and you end up with 5.95V DC when loaded with 0.5 to 1 Amp, that is close eneough to the 6.3 V spec. This worked well for me a number of times.

Schottkys are available in normal diode packages ( examples: 1N5822 for 3 Amps or STP1545F for 15 Amps) but cheapest way is often to use the 3 legged TO-220 Transistor looking ones with 2 diodes in it with a common cathode connection, one of these forms half the bridge, add 2 more for the other half of the bridge and in each of those either use just one of the diodes in the package or connect the 2 diodes in parallel by connecting the two anodes together.

Cheers,
Ian