Bias Supply Filtering

[ChrisM]

I see some amps filter the bias supply voltage after the diode (rectification) and also on the output of the bias. However most amps leave off that extra filtering on the output. What's the deal with the extra filtering, is it needed, advantages, tonal affects?

Also how does the size of the caps affect the tone of the amp?


Thanks...

[breakfastbuddy]

the connection is taken from ac . so it need that filtering after the diode

[ChrisM]

Ya I understand that.

I am asking why some amps have this added filtering on the bias supplies output.

[paulster]

It's mostly down to designer preference, and which amp they originally used for 'inspiration'!

Be careful arbitrarily upping bias filter capacitor values though. If you look at the ramp-up plots in a SPICE simulation you can see that a seemingly innocuous swap like changing a 22uF for a 100uF cap can cause the bias voltage to take over 60s to come up (down actually) to full bias voltage at power-on.

Do this and flip the standby within this time and you're running with next to no bias.

[ChrisM]

Hmm interesting... I know the ODS uses a 100mF to filter.

There must be a tonal effect or reliability factor to the cap size though? Is there wasn't wouldn't everyone use 22mF.

[Bob-I]

Since the bias supply connects directly to the grids of the power tubes, any small noise is directly amplified by the output tubes. Some amps just cheep out on the filtering there, not a good idea IMHO. The extra filtering costs a few bucks and really helps.

[Structo]

Also when talking about relibility, there are different ways to wire the adjustment pot on a fixed bias amp.

I haven't utilized this but it makes sense that you don't want the amp to run away if the pot fails.

[David Root]

As I understand it it's about getting capacitance localized near each node, same as in a conventional PS supply. There it prevents motorboating, in a bias circuit in can prevent hum coming in on the bias supply.

Also I would use 220K 1/4W, or 1/8W if I could find them, rather than 1M, but the main point is right, it's a bias failsafe item that I need to put on both my D-amps. I use smaller bias pots and 1/2W simply don't fit, too big.

[ChrisM]

Thanks guys,
I am going to use two 47mF 100V caps then. After the diode and on the output.

Tom, I am going to wire Fail Hot because that's how all my amps are and I can't really see my bias pot failing. It has such little DC on it.

[tubeswell]

Thanks guys,
I am going to use two 47mF 100V caps then. After the diode and on the output.

I use a Pii filter with 2 x 47uF 100V in my bias supplies and they work fine

[David Root]

Me too, with same result.

[ChrisM]

What inductor's are you using, size and current rating please?

[PCollen]

Ya I understand that.

I am asking why some amps have this added filtering on the bias supplies output.

Good info on bias circuits:

http://www.el34world.com/charts/Biascircuits.htm

[martin manning]

In addition to providing some additional filtering, the extra cap on the output bypasses the bias divider and insures that the power tube grids see the same ac impedance to ground regardless of where the bias pot is set.

A little ripple in the bias supply shouldn't be a problem though, as a push-pull output stage will eliminate it through common mode cancellation. And, if the bias filter caps are big they will take a long time to charge, and it will be some time before the bias voltage works its way down. Using 100uF's will result in nearly a minute going by before the voltage stabilizes, and this could mean running under-biased for a bit at start-up.

MPM

[PCollen]

In addition to providing some additional filtering, the extra cap on the output bypasses the bias divider and insures that the power tube grids see the same ac impedance to ground regardless of where the bias pot is set.

A little ripple in the bias supply shouldn't be a problem though, as a push-pull output stage will eliminate it through common mode cancellation. And, if the bias filter caps are big they will take a long time to charge, and it will be some time before the bias voltage works its way down. Using 100uF's will result in nearly a minute going by before the voltage stabilizes, and this could mean running under-biased for a bit at start-up.

MPM

Good point, Martin. With 100K preceeding a 100uF cap in a simple 'Dumble-like" bias circuit with bias derived off a dedicated low VAC secondary winding, it would take about 47 seconds (5 RC time constants) for the cap to charge up to 99%. Fortunately, that resistance is typically 1 to 10K (the bias range resistor), and 99% charge typically occurs in roughly 0.5 - 5 seconds.