Tube power supply design, would you have a look?

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

plumsolly

Well-known member
Joined
Nov 3, 2007
Messages
969
Location
Denver, Colorado
I am putting together a power supply for a pair of Cinema Engineering 7080 valve eqs (see here: http://www.groupdiy.com/index.php?topic=47004.0)

Power requirements for two units is 300 volts dc at 48 ma and 6.3 volts dc at 1.8 amp.

I had a nice UTC R-112 PT (350-0-350V@120ma, 5V@3A, and 6.3V@5A) laying around, so I designed it around that. 

I made it passive, for simplicity.

I made the HV tube rectified because I could, and because I have the voltage to waste. I did not use a choke on the HV because, again, I have the voltage to waste and because I don't have an appropriate one.

I made the heater supply a voltage doubler so I would have a little room to maneuver with the filtering, and because I have the current available. I did not use a choke for no other reasons than that I don't have one and finding a 2A choke is tough. I don't really have much voltage to work with here for the filtering, so I guess I might need to find one if the RC filtering is not sufficient.

A couple of things I am not sure about:

The value of those first two caps in the in heater supply.
The current rating of the diodes in the heater supply.
The current rating of the rheostat in the HV supply. (It is passing two amps of DC, but what about ripple? Do I treat that separately?)
Do I need to worry about regulation at all?

Thanks for your help!

Best,

Ben
 

Attachments

  • Cienma Power Supply.jpg
    Cienma Power Supply.jpg
    153.4 KB · Views: 113
The HT side looks OK to me. It is a line level audio circuit so you don't need to get HT ripple down to the microvolt level.

The heater supply is a bit odd. It looks like you voltage double the 6.3V ac then smooth and drop it back down to 6.3V dc. Although you have plenty of volts to lose it is very energy/heat wasteful and requires large wattage resistors and the smoothing operates a line frequency. Again, as this supplies a line level audio device you probably do not need the smoothing  on the heaters that you would need if it were a mic pre. Just to put this into context, if you just used 6.3VAC the mains hum peak to peak voltage across the heaters is 17.8V. It is unlikely that the hum this produces at the output of a line level circuit would be greater than -60dBV (or one millivolt in normal money). If you could reduce this by 30dB it would be negligible which means you want it to be 563mV peak to peak.

OK, suppose you full wave rectify the 6.3V and slap a big cap across it. How big does that cap need to be to get this ripple with a 2A current draw??

We use that well known formula C dV = I dt rearranged so that C = I dt/dV

dV is the peak to peak ripple we want i.e. 0.563V; I is the load current 2 amps and dt is the time interval which for 50Hz mains and full wave rectification is 10mS. According to my calculator the value of cap required is just under 36,000uF. Your existing design already uses 80,000uF so a full wave rectifier with these caps and a 2 amp load would have a ripple of  only 0.25 volts or 37dB below the raw ac value. I would say that was more than adequate.

Some people think the rectifier diodes in this circuit need to carry huge forward current spikes but the transformer secondary resistance and the diode forward resistance usually limit these to ten times the output current or 20 amps in this case. A standard 1N5401 3amp diode is rated at ten times this amount.

The other thing you might like to consider is elevating the heater voltage to 50V or so. The reason for doing this is to ensure the heater is always positive with respect to the cathode so there can be no electron flow from the heater to the cathode.

Cheers

Ian
 
Can you even get reostats in those values and powers?

Don't put the reostat between rect and first cap. Yes, the RMS heating is much more than just the DC, and the difference is a nasty computation.

Don't use reostats. Get a 10-pack of 1K 5W resistors and use as many as needed.

Doesn't have to be 300.00V. The amp will work at 270V or 330V, with mild differences in headroom.

That heater supply. You are doubling-up to 18V but using 6V. Total DC load is 18V 2A or 36 Watts. Winding is rated 6V 5A or 30 Watts. There's no free lunch. The winding is 20% over-loaded.

Also the 18V-6V or 12V drop in dropping resistors will be less if a tube is pulled for diagnosis; perhaps putting 9V on the remaining tube(s). That's not quickly fatal but is rude to the tube.

Agree that 6.3V AC never gives a happy 6V DC.

Can you wire your heaters series 12V? Then you need 12.6V at 0.9V, which is easily had from 6V 5A doubled.

And a series-string with one tube pulled goes dead, not un-sag to a higher voltage. (Anyway you won't need 2/3rd sag, maybe only 20%, less up-side risk.)

_I'd_ go the whole way. Lose the hollow-state rectyfire. Strap 6V and 5V windings for 11VAC, doubles to 29VDC, can feed a 24V string. (Ah, 3 bottles per unit, a 19VDC 0.3A goal seems best.)

And higher-volt series-string allows you to put the sensitive cathodes "elevated heater" as Ian suggests, though note that in the Cimema several cathodes are already up at 8V so "elevation" does not happen until 10V-12V up the string (put the push-pull stage at the groundy end).

Again, I'd get a 10-pack of 1r-10r 5W/10W resistors and use as many as needed.

The diodes should be rated for the _winding_, with a generous margin. 5A winding needs 5A diodes at least. The added cost of 10A parts is less than the added cost of replacing 5A diodes at an unsuspected moment. (And I'd use 400V diodes because line spikes do happen and plain Silicon isn't more expensive in DIY economics.)

Capacitors should be >1000uFd/Amp, much greater with today's low-cost low-volt caps. Assuming 29V 0.3A, >300uFd. The first doubler caps could be 2,200uFd 25V, the next two may as well be 4,700u 35V.

Ripple is then 0.2V at first caps, about 50:1 reduction of 100/120Hz in each RC, 0.1mV ripple on the heater line. Don't wrap it around first grid; otherwise this is plenty clean.
 
RCA put the rheostat between the HV center tap and ground in their 50 mA preamp PSU. 

The Collins PSU for the entire 212E/F/G consoles goes cap/choke/rheostat/cap.  It is rated 250-300 @ 250mA, and uses a 2K5 50W rheostat.  I had to replace one, and it was about $40 from Mouser. 

I agree you don't need the rheostat, just dial it in on the bench.  It's not a value that will have to change, like with the RCA and Collins systems.   

I have some spare NOS/NIB 8K/50W rheostats if you want to go there, same as RCA used in one version of the preamp PSU. 
 
High Voltage Section: go choke input.

don't have a choke?  do you have any other power transformers laying around?

like a 120/6.3 ?

use the 120 pri as the choke, cut off the green wires,

this will knock your over rated transformer down with no heat,

use reactance, that j operator thing with the 90 degree shift,

6.3 section is a train wreck,

diodes too small, no need for doubler,

go for a huge 25 amp bridge,

the square ones with thick leads and a hole in the center, with the metal backing that gets white greased to the chassis for a bullet proof circuit,

see, every time you turn on a tube circuit, you have a dead short to the 6.3 pwr supply in the form of cold heaters in parallel,

inrush can be brutal, so you need a huge safety margin for heater circuits,

the 25 amp bridge at 1000 volts will be about 89 cents at the surplus if you live close,

you do not need perfect DC for heaters, the tubes are lucky just to get DC at all,



 
You might consider solid-state HV rectification, and stacking the 5V and 6.3V windings; recitfy that and you get (depending on diode drops) about 13.5 V, which you could lower to 12.6V with a dropping resistor..

Or you might try this: build the HV supply, make a voltage divider to create about +75V, then connect that to the center tap of the 6.3V winding. Run the heaters on AC with this elevated filament winding and I bet you'll get plenty quiet results. NY Dave did that on one of his mic preamp designs, and your signal voltages will be a lot higher.

Peace,
Paul
 
pstamler said:
Or you might try this: build the HV supply, make a voltage divider to create about +75V, then connect that to the center tap of the 6.3V winding. Run the heaters on AC with this elevated filament winding and I bet you'll get plenty quiet results. NY Dave did that on one of his mic preamp designs, and your signal voltages will be a lot higher.

Peace,
Paul
I just finished the NYD MILA with ac heaters and above addition. With careful layout you should have no noise.  And I second a choke.
 
Maybe use one of the Intl Powers ?  I think all you have to do is supply a housing and add your own filament.  Already comes with the main PT and is regulated.


http://internationalpower.com/cgi-bin/htmlos.cgi/002.1.044762400515529893
 
So long as you can find the specs you need, it's hard to beat the international power supplies. It's built, only needs wiring and fusing, gives you UL / CE / TUV approval/protection.  All for about $45 from Mouser. 

http://www.internationalpower.com/generalspecs.html
 
Thanks guys. Unfortunately, those international power supplies only go up to 250V. I went ahead with Glassware supply kit and an Antek PT and its working great. I have a couple of questions though.

Is anyone a fan of fusing the heater and B+ lines individually (in addition to the mains)?

I need to power a couple of leds and a couple of relays and was hoping to borrow some 12.6vdc from the heaters for this purpose. Is there anything inherently wrong with that? One issue would be that the PS board is set up to elevate the heaters by 1/4 of the B+ voltage (75V in my case), so I would have to use the "heater ground" 75V at the relays and leds. Or de-elevate the heaters which would probably be ok as there are dc anyway?

Thanks,

Ben
 
Back
Top