My Official Newb PSU Questions - Filters & Transformers

[northsiderap]

I am doing a simple transformer isolated 2-stage mic preamp using a 7247 and a 12AU7.

This is my first tube project, so I have a question or two before I finish the wiring.

I have 6.3V AC available. How do I wire up the heaters? In through the C.T. and the 'H' connection to ground? I really don't know what 'series' and 'parallel' really mean for heaters. Do these terms refer to more than one tube in series or parallel, or do they refer to the way you use the heater center tap?

2) How does one determine the resistor value from the filament supply to ground in the PSU? Why is this thing there anyway?

Someone better explain it before I guess and make a fool of myself.

Parodon my newbness when it comes to tubes, and thanks.

 

[CJ]

Yes, either way on the 12au7, ground pin 9, pins 4,5 tied together and then 6.3.

Or tie 4 and 5 , ground them , and then 6.3 goes to pin 9.

Two 100 ohms to ground form a virtual center tap, which can reduce hum.

 

[northsiderap]

Thanks...

So 'series' is NOT using the c.t. and 'parallel' is from center tap to both heater taps.

 

[adamasd]

In a 12AU7 pins 4 and 5 are the heater with pin 9 being a center tap, consider them two 6 volt heaters connecting at pin 9. If you wire them in series one side of the AC goes in pin 4 and pin 5 gets the other side of the winding, so they split the available voltage. Feed them 12 and they each get get 6, which is their rated voltage. If you bridge pins 4 and 5 and give them one side of the winding and give the other side to pin 9 they are in parallel, so they each take the same amount of voltage, so 6 volts.

I always just use 100 ohms, most commercial gear uses 100 ohms so I leave it at that. There was a thread about this not to long ago and there was some discussion about calculating a value, but I think the conclusion was its not that critical, 100 ohm to 1k, just make them the same and as close as possible, or a pot instead so you can balance the resistance seen on both sides.

adam

 

[northsiderap]

More Questions:

More questions as I read about the subject, and help from any of the local gurus is always appreciated.



1) How much do I load the secondary of the mains transformer? Do I want to actually calculate the turns ratio and use the first resistor after the rectifier to reflect a specific load to the primary, then find a capacitor to build an appropriate LPF?

2) When looking up low-pass condenser filters, I've noticed that the knee is often @ 1Hz. Isn't this a slow response time for the initaial stage, which is supposed smooth out a 60-120Hz pulse?

For Reference:

 

[Guest]

I don't think that much precission is required. You know the current draw and the voltage the secondary supplies so you can calculate an appropriate resistor or choke for the PI filter. Less ripple is obtaines if you use multiple sections. Capacitors are usually chosen for what fits and what it costs. As long as it is reasonably large (47uF-470u)F it will work fine.

A very helpful program for looking at power supply parameters is here:http://www.duncanamps.com/psud2/index.html It is in the metas as well.

 

[northsiderap]

I have built this 220V PSU from a 250V @20mA bridge rectified transformer.

I am using 10W resistors for a three-stage RC filtered power supply. I am doubling the wattage to battle the inital voltage spike and the Wattage AC combined with the wattage DC for the initial stage. From studying other designs, this should be more than enough. But...

During the inital turn-on, PSU Designer II calculates my wattage spiking at 68 Watts for the stage 1 resistor, 1/2 watt at my second stage resistor, and 1 Watt for resistor 3. I don't get it.

[edit] I might be entering the simulation source transformer resistance wrong - - - As the source resistance goes up the wattage on stage 1 R goes down.

These are the values I'm using in order as looking out from the bridge:

BRIDGE
R1: 2k
C1: 47
R2: 220
C2: 47
R3: 3k
C3: 25
B+ OUT

 

[Guest]

I am not sure why you are using the values you chose, but in general, using a resistor or choke right after the bridge stiffens the supply a bit. If you just want to get 220VDC from your 250V transformer at 20mA you could do a number of things but here is one:

The transformer output from the bridge is 250Vx1.4=350V
You need to drop 350-220=130V
130V@20mA gets about 6.5k

If you use a standard PI filter you could use three 47uF capacitors and two 3.3k 5W resistors and get a smooth 220VDC B+ (CRCRC).

As for the PSU designer, if you enter the transformer data in the appropriate box (ex. 250V, 100mA, 10%) it gives pretty good results. I don't think an initial spike at turn on indicated by the program is accurate unless your data is accurate.

 

[northsiderap]

I am using those resistor values because they are what I have available to me right now. All of this is built from an old VCR and tube TV set.

From my recent reading, I understood that it was wise to at least double your resistor wattage to take into account the initial voltage spike and combined wattage of both DC and AC for at least the first stage.

What the text didn't explain is perhaps I am only figuring out the wattage for the voltage DROP across the resistors? That does make more sense than figuring out the wattage through the entire filter for the B+ voltage.

I had decided to do a RCRCRC filter instead of the brute capacitor-after-bridge supply because I am anticipating a 10V output swing on the plate, and I have read that a resistor rather than a capacitor after the bridge is slightly more stable.

You have cleared up a couple of things for me though - I have a habit of making things too complex.

I am running two 5 watt resistors in [edit] parallel for R2 - - If I took one of the 5watt parallel R off of that stage it would be 3.5k 5watt.

 

[Guest]

Wait a second. You list R2 as 220 ohm. If it is 2 resistors in series, taking one out would have to make the resistance less, not more. Do you mean parallel?

For a three stage PI you will get more than enough ripple rejection without the R1 right after the bridge. If you use it, just remember to take into account the voltage drop so you get the right B+.

If you use R1,R2,R3 at 2k, 3k, 3k and run 20mA you drop 160V

 

[northsiderap]

[quote author="thomasholley"]
For a three stage PI you will get more than enough ripple rejection without the R1 right after the bridge. If you use it, just remember to take into account the voltage drop so you get the right B+.

If you use R1,R2,R3 at 2k, 3k, 3k and run 20mA you drop 160V[/quote]

Some texts would take six pages to say just this. So far so good. Thanks!

 

[PRR]

> low-pass condenser filters, I've noticed that the knee is often @ 1Hz. Isn't this a slow response time for the initaial stage, which is supposed smooth out a 60-120Hz pulse?

It is supposed to REALLY smooth a 120Hz pulsewave.

If you set the low-pass to 120Hz, it would knock-down the ~320V ripple just 3dB, to ~200V, which is still mighty huge.

If you want to knock it down a lot, with a single pole filter, you set the filter frequency WAY below the undesired frequency. Traditionally 10-20 times lower. With cheap modern electrolytics, 100 times lower is not expensive.

> Do I want to actually calculate the turns ratio and use the first resistor after the rectifier to reflect....

Huh? The transformer sees the first CAP, through the rectifier, shunted by the TOTAL DC load. The cap soon charges up, so the total DC load is the working load. Here we have a note "320V 13mA", so we know the total of the series resistors, the Zeners, and the tubes is 320V/13mA= 25K ohms. If the load were 50 ohms, and dirty power were acceptable, 1,000uFd would "work". Since the load is 25,000 ohms, we could go 2uFd. But we need very-clean power, so we increase it a LOT. At today's cap prices, 100 times bigger is not a wallet breaker.

> During the inital turn-on, PSU Designer II calculates my wattage spiking at 68 Watts

If the transformer has low-low resistance, the first 220uFd cap will charge to almost 320V on the first half-wave. The second 220uFd cap is still at zero volts. The first 3.3K resistor has almost 320 volts across it. Power dissipation is 320V^2/3,300= 31 watts (I dunno why you get 68W). However it will charge-up to 63% or 202V in about 1 second. Now it has (320V-202V)^2/3,300= 4 Watts. If it ultimately goes to 13mA, it will settle down to 0.56W.

Can a 1 watt resistor take 31 watts for one second? Hmmm. I've seen similar things fail after a few years, and indeed it is beyond the rating of many resistors. While caps are cheap, it might be about as cheap to use more R-C stages with smaller caps, so we didn't have to use 5W 10W resistors for a watt of operating heat.

> BRIDGE R1: 2k C1: 47

Ah, now I see your 68W, well, at least 51 watts.

If all you got is junkpile parts, just throw it all together and see what smokes. Power supply calculations are mind-benders, and often flawed. Reality is the best reality-check.

 

[northsiderap]

It worked! I put the cap before the resistor... @380VDC unloaded. It didn't take too long to ramp up to 380, then do the slooow climb as the caps continued to charge. A little more than I expected. :? I'm going to add a bleeder in there - My caps are only 400V which is too close for comfort IMHO. I guess I'll try for less than 10% on the bleeder, since my tubes are going to rob me of @15mA

If I let it sit for some minutes, it likes to creep back down to @230V and sit there. Any tips on discharging caps? I'm probably going to attempt a shielded scredriver with a ground clip as I believe Larrchild posted somewhere.

(approximate values)
250V Secondary
Bridge
47uF 450V
2k
47uF 450V
3k
25uF 400V
B+

After sweating it for a bit, I plugged it in and it runs as cool as a cucumber.

Then I run around the house shrieking like a schoolgirl.

 

[Guest]

Congratulations. I recommend you should install a permanent bleeder of say 100k or so to discharge the capacitors slowly at turn off. As a discharge tool I usually use a low ohm power resistor with clip wires. The screw driver works but I am clumsy and usually shock myself. I don't like the big sparks either but I am a sissy.

 

[northsiderap]

Since the coupling caps I have available are all 250V rated, I'm going to have to add a voltage divider.

I'm pretending the filter resistors don't count, I'm not sure if they do or not. The tubes are constant current sources.

Any second opinions? Am I going about this the right way at all?

Most likely the real-world values will be:

R3: 20k
R2: 10k
R1: 60+k

 

[Guest]

I'm not exactly sure what you are depicting but I will make some guesses. You start with 380VDC unloaded. When you load the supply with your tube circuit, drawing a presumed 20mA you will drop about 100V (3k+2k*20mA=100V) so you start with 280VDC supply. If you are only drawing the 9mA in your graphic, then you would have a 45V drop (supply would be 335VDC).

You can adjust this voltage lower for each stage using the divider topology of resistor in series and decoupling capacitor to ground with the highest voltage at the output and lowest voltage at the input.

A bleeder resistor is generally used only in the power supply, in parallel with the last capacitor, in your case the 25uF 400V one.

 

[northsiderap]

I was doing this:

http://www.tpub.com/neets/book1/chapter3/1-35.htm

See the next few pages also.

 

[Guest]

Yes, I see the voltage divider, I am just not sure why you are doing it like this. It will work fine I think, but just not "traditional" way of doing a tube power supply.

As I mentioned above, the loaded voltage of your power supply will be somewhat less than the 380V you have unloaded. The amount of voltage drop depends on the current draw so at 20mA you have a supply of 280V to your preamp. Generally, you should decouple this supply voltage at each stage, in similar fashion to the voltage divider you showed, except after the series resistor, you place a capacitor to ground to decouple power supply ripple from that stage. You do this decoupling physically near the tube for best effect. You usually do it between stages, beginning from the output and proceeding to the input, which has the lowest voltage. If you need lower voltage from the supply to your output tube you can decouple it there as well.

The supply bleeder is generally placed at the supply and not in the preamp. I may be wrong, but I think if you put a bleeder in the preamp first stage it would contribute noise to the stage. Even if it isn't noisy, it is not necessary so you save a penny.

Some good examples are the power supply circuits NYD has posted.