Caps in parallel and series in PSU, experiences?

[utervo]

I am doing a triple build with two Deluxe Reverbs and a Super Reverb.

I have a bit of a hard time finding suitable electrolytics for the power supply. Especially the high voltages on Super Reverb are a bit of a concern if the amp is ever to be switched on unloaded.

I want to stay close original values and not definitely not stiffen the amps.

I can get 350V or 450V F+T caps here with a good price, if I use 33uFs I could do parallel 33uF caps in series with another parallel 33uF caps to do a high voltage 33uF first section (35uF in SR and 32uF in DR originally).

The next sections would be 33uF caps in series, so high voltage 16.5uF sections (20uF originally in SR and 16uF in DR), I could measure the caps and use the higher ones for SR and lower ones for DRs.

220K bleeder resistor parallel to caps, counting paralled first section caps as one cap if I understand correctly, so eight resistors in total.

I think this should work and the caps would last longer staying at a lower temperature, price also wouldn't be too bad compared trying to find correct size caps at different places.

What kind of experiences have you got with putting power supply caps in parallel and series? Any effects on tone and feel if total capacitance stays same?

[ToneMerc]

220K bleeder resistor parallel to caps, counting paralled first section caps as one cap if I understand correctly, so eight resistors in total.

I think this should work and the caps would last longer staying at a lower temperature, price also wouldn't be too bad compared trying to find correct size caps at different places.

What kind of experiences have you got with putting power supply caps in parallel and series? Any effects on tone and feel if total capacitance stays same?

Here's how I do the first cap bank, here's two 80uf caps.

https://tubeamparchive.com/download/file.php?id=24406





TM

[utervo]

I have a bit of hard time seeing what is happening there. The second cap is not soldered to the turret from negative and the series connector wire is under the board?

[utervo]

Received some specs from Classictone, so the voltages would be using 230V:

Deluxe Reverb
337.6-0-337.6V

Super Reverb
353.9-0-353.9V

So rectified maximum voltages would be:

Deluxe Reverb
337.6 x 1.414 = 477V

Super Reverb
353.9 x 1.414 = 500V

For DR I could have gotten away with 500V caps I understand. In your experience does series caps make a difference in tone and feel if capacitance is equal?

[gingertube]

You MAYBE could have got away with a 500V cap except on the day the mains voltage had a surge of 10% high.

When re-capping Fenders I always use a series connection designed to give the original capacitance value, that means series connection of 2 caps of double the value. (Or parallel 2 lots of series connected caps of the original value).

Next - don't get too hung up on getting the exact same capacitance. Depending on the brand, the value will be specified as +/- 20% OR -10% + 30% tolerance. So series connecting 2 of 47uF to replace a 22uF etc. is fine.

Next - Capacitors share voltage according to their capacitance, this is an inverse relationship. That is, if you series connect a 50uF and a 100uF then the 50UF will end up with twice the voltage across it as the 100uF (the same current flows through them so the smaller cap charges to a larger voltage).
SO always fit voltage share resistors across the caps. Many forget these and many who do remember to fit them, use way too large a resistance value. What you want is to have 3 to 5 times the current flowing in the voltage share resistors as the leakage current of the cap.
Most caps will show the leakage current on the datasheet - if not calaculate it from this default formula:

0.006CV where C is the capacitance value and V is the applies voltage (not the working voltage of the cap). This is appropriate for most high voltage capacitors

Another "default" formula you will some times see for low voltage electrolytics (Say 63V or less) is capacitor leakage current is 3 x squareroot (CV) microamps.

Example:
500V and 100uF
Series connect 2 off 220uF/350V Capacitors.
Each Cap will have (hopefully) 250V across ir
Capacitor Leakage Current will be 0.006 x 220uF x 250 = 330 microamps
So want around 1 to 1.5 ma through the voltage share resistors - use 220K across each cap to give 1.14 mA. Power in the resistor (I squared R) will be 0.284 watts, use a 1 Watt resistor minimum so that it doesn't get too hot therefore help to dry out the electrolytic cap and cause early death.

Changing capacitor values within a +/- 100% (that is, half to double the original value) won't make a lot of sonic difference. You need to make a bigger change than that to have a noticable change in sound.

Cheers,
Ian

[utervo]

Thanks for the tips Ian.

I am planning to use 220K resistors to balance the voltages as you suggested.

Just waiting for parts to arrive in the mail.....

[martin manning]

Well done Ian, thanks!

Utervo, per Ian's calculation, 220k balance resistors are appropriate for 220uF caps with 250V across them. Since yours are only 33uF, you could use 220/33 x 220k = 1.5M resistors. You could go at least to 470k and reduce the load on your power supply by half.

Power in the resistor (I squared R) will be 0.284 watts, use a 1 Watt resistor minimum so that it doesn't get too hot therefore help to dry out the electrolytic cap and cause early death.

Note that the same amount of heat is released from the resistor regardless of its wattage rating, but due its larger size the surface temperature of a 1W resistor will be lower than that of a 1/2 W. I would think that as long as the resistors are some small distance away there will be little difference in the operating temperature of the caps.

[utervo]

Good point Martin, did not think of that, I have the 220K's already ordered.

Going by the original voltages the extra strain would be:

Super Reverb:
465/(220+220)=1.06mA (plate, in the original, not extra strain)
460/(220+220)=1.05mA (screen)
400/(220+220)=0.91mA (guessing voltage here)
270/(220+220)=0.61mA (preamps)

So extra guessed strain compared to stock would be 1.05+0.91+0.61=2.57mA, does that sound like much?

Deluxe Reverb:
420/(220+220)=0.95mA (plate)
415/(220+220)=0.94mA (screen)
250/(220+220)=0.57mA (guessing voltage here)
180/(220+220)=0.41mA (preamps)

So extra guessed strain compared to stock would be 0.95+0.94+0.57+0.41=2.87mA, does that sound like much?

[martin manning]

I get 3.63 mA for the SR, and 1.5W of heat for the balance resistors. We agree on the DR at 2.88 mA, and that's 1W of heat. Not a big deal, only about 1% increase in current.

[utervo]

Yes I left that first one off from Super Reverb as it is there on stock, I only put the extra strain on PT there. Maybe I could use 470Ks then. I have to think a little. Ian suggested 3-5 times leakage current, Valve Wizard also suggests at least 5 times the leakage current. I will look at the specs of the caps and see what the leakage current is for A-series caps.

[utervo]

Datasheet says leakage current is:

≤ 0,008 * C[µF] * U[V] + 6 µA [µA]

But what do I put for voltage and capacitance if the caps are in series...?

[utervo]

Does anyone know what the original Deluxe Reverb and Super Reverb PTs are rated at?

[gingertube]

utervo,
They certainly mean the value of the capacitor but I think they mean applied voltage and not the actual voltage rating of the capacitor.
BUT
Since the maximum applied voltage should never exceed the voltage rating then using the capacitors voltage rating instead should give you a "worst case" leakage current value.

It is actually good engineerining design practice to design for worst case conditions - so using the actual capaitors voltage rating in the formula is a GOOD idea.

Cheers,
Ian

[utervo]

deleted

[utervo]

But then again that doesn't take account the +-20 tolerance on cap sizes... I have a feeling (too tired to have anything more solid anymore!) that 220K for plates and 470K for rest will be good. Must continue tomorrow.