SVT re-cap - how much can I increase the capacitance values

[xm52]

I am hoping that I can get some advice on capacitor values for an early 70's SVT. Here's the schematic: http://www.el34world.com/charts/Schematics/files/ampeg/Ampeg_SVT_PowerAmp_RevF.pdf.

I've always found that the power supplies in a number of Ampeg designs can benefit from increasing the capacitance, especially in the first node. Whenever I've done this, it has worked out well in terms of better low end, increased headroom, and reduced hum.

These are the original values and the changes that I'm thinking of trying.
Node A: 52uF/900V (70uF/450 || 40uF/450) + 100uF/450      ->  ????
Node B: 30/600V (60uF/300 + 60uF/300)                              -> 100uF/450V + 100uF/450V = 50uF/900V
Node C: 40uF/450V                                                              -> 100uF/450V + 100uF/450V = 50uF/900V
Node D: 90uF/200V                                                              -> 100uF/350V
Node E: 100uF/450V                                                             -> 100uF/450V                                        
Node H: 40uF/450V                                                              -> 50uF/450V

I think that the slight boosts in nodes B through H are reasonable.

My concern is what to do with node A. Normally when recapping I use a 2X100uF/500V cap can connected in series here. I have a couple of 270uF/450V Panasonic caps which in series would give me 135uF. Considerably more than the original 52uF. My concern is the strain that this would put on an old power transformer.

I find the you can increase the value to improve the performance but above a certain level, it doesn't make that much of a difference. Does anyone have any insight on what a good target value for Node A should be?

Thanks.

[HotBluePlates]

Are you trying to avoid the use of can caps? What form factor of cap will you be using to replace the originals?

Why is node C changing from 450v rated to 900v rated?

[firemedic]

I think he's just trying to get the uF value closer to original spec on the series 100's, using what he has.
I personally would go with the 135uF value, that's not really that high with a SS rectifier. 

[xm52]

I'm not trying to avoid using cap cans. The amps has two type-FP cans. If I use a cap can, I would want to limit the choice to that make. I wouldn't want to drill a hole in the chassis for a mounting ring if a different type were used. I also want the top of the chassis to retain the original look.

When I recap these amps I usually use two type-FP cans and three axial tubulars in the power amp. The values remain close the the originals. In this case, I want to experiment with a different make of cap and boost the Node A value.

I intend on either building a circuit board for the caps and mounting it under the chassis, or putting pairs of caps into the existing cap cans. The problem with a circuit board is that space is tight under the chassis. There are three axial tubular caps there now and I'd like the board to be mounted where they are located. There are a number of wires in the way in that area so I'm not sure if there is enough space to mount the board. The other alternative is to take cap pairs and mount them in a tube.

I have a supply of Panasonic TS and HC series caps that I intend on using. They are PC board snap-in capacitor. I want to see how an SVT sounds with them. I have been using these under the chassis in my builds and they perform very well wherever I've used them. Sometimes I use a Dynaco MKIII replacement board (SDS Labs) from Triode Electronics. This board is too large to fit under the SVT chassis.

There is no special reason for boosting the Node C voltage from 450 to 900 other than I have the caps on hand and they would match the Node B pair on the circuit board.

[PRR]

> Why is node C changing from 450v rated to 900v rated?

Because at turn-on node C will rise to 700V, dropping below 400V after several seconds. The 450V cap will stand this long enough to expire the warranty, but it dies over time.

Nodes A B C should be 700V rated. With electrolytics, this means at least two 400V parts in series. 450V parts (900V per stack) are common.

I feel node A should be ~~120uFd not 50-some uFd, that cap-costs hurt Ampeg's profits but we can take a more long-range view now. Two 200u-300u seems reasonable. No, it will not hurt the PT.

[xm52]

Thanks for your replies.

Ampeg must have been getting a good price on those 70/40/40 450V cap cans. They used them in several of their designs. Even in cases like the SVT where they should have been spending a little more to put in more capacitance.

Now that I'm planning on changing from cap cans to Panasonic TS caps in this amp, I realized that I should consider the ripple current spec. The TS-EE cap (http://industrial.panasonic.com/www-ctlg/ctlg/qABA0000_WW.html 220uF 400V has a 1.79A_RMS ripple current rating at 120Hz and 105C. I've used mostly HA and HC series caps which have a lower ripple current rating. I've never seen a spec for the original Mallory FP cap cans or the ones that CE now manufacturers on the old Mallory equipment. I'm looking into that now. It would be interesting to see the cap can specs.

[PRR]

This is a BIG brutal amp and it is good to wonder what the stress is.

I doubt Ampeg calculated it much.

As you say, it appears they had a truck-load of these caps, they worked in one amp, Ampeg used them in everything, and the stuff worked (didn't come back for a lot of warranty repair).

Three pair 6550 worked HARD:

diodes:
700V peak (you should double this for wall-surge)
1.01A RMS (a 1A diode is not enough!)

main cap:
3.4A peak 1.3A RMS

> 1.79A RMS ripple current rating at 120Hz and 105C

This should be quite acceptable. 1.8 is higher than the 1.3 we need. The rating is at 105 C, and I hope you won't boil spit on your caps (they will be HOT but not THAT hot). Also the 1.3A happens only at FULL ROAR. At idle it is more like 0.6A RMS. In typical LOUD use, somewhere between, 1.0A-1.1A.

Taking 60 deg C cap-box temp and running around 2/3rd of rated ripple, you could predict(*) about 90,000 hours MTBF. Playing 4 hours a night 6 nights a week, that's 77 years. In fact they will die of SOME obscure cause in 20 to 50 years. And any manufactured part may die at any time. But odds are very good your fingers will stiffen-up before your caps dry-out. (Yes, Les Paul... but even he never used an amp un-repaired for 77 years.) It can be your heirs' problem.

(*)Actually Psonic says the formula for TS-EE is different from other caps, contact them. If you calculated 99 days and offered a 90-day warranty, and were selling heaps, you'd want to get the detailed computation before you drowned in unexpected 88-day warranty returns. For a one-off with rough-estimate similar to owner's lifetime, especially for owner doing his own "warranty work", I think you can just wing it.

[xm52]

Those numbers are very interesting!

In particular, the original diodes were 1N4007's. They were being pushed pretty hard. I've substituted 3A (UF5408) diodes.

It looks like it would be a good idea to go with the higher current capacity TS-EE caps as opposed to the TS-HC's (1.1 A_RMS) that I initially thought I might be able to get by with.

Thanks again.