1n400x for gainclone?

can i use 1n4004 for double bridge psu for lm3886?
Rise the bet to 1N5402 or higher.

1n400x series seem desirable for me since they're 'soft recovery' types
What does that mean?
or to be more precise: what is it supposed to "improve"?

would the chip become current starved?
No, if you demand more than 1 A from them they will happily supply it, dying in the process.


Paid Member
2009-02-11 12:58 am
Rise the bet to 1N5402 or higher.


The recovery characteristics of a diode refers to what happens when the diode goes from being forward biased (conducting) to reverse biased (turned off). The diode doesn't turn off completely. There are two main specs here: The speed of recovery, slow/fast, and the recovery profile, soft/hard. It all relates to how the carriers move out of the PN junction when the diode goes into reverse bias. The idea with a soft recovery diode is that the carriers take a little bit to get out of the junction and they move out in a smooth fashion. This minimizes the discontinuities in the diode current and, in theory, sets off less ringing in the transformer secondary.

Rather than four discrete diodes, I suggest looking at a bridge rectifier. Something like a GBPC2504 would be perfect and plenty rugged. Remember, the diodes also need to survive the inrush current.


Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
For LM3886 chip amp power supplies, I go hog wild and use avalanche-rated, 30 ampere, discrete diodes with datasheet guaranteed soft recovery characteristics: (datasheet). Their inrush current rating is 300A, way more than an Antek 200VA power transformer is able to source, even when you place a dead short across all filter capacitors.

Using these diodes does cost a bit more money (4 x $1.52, versus 1 x $1.80) but you end up getting some of that back, when you buy four individual HS368 heatsinks instead of one big ole bohunker HS335 heatsink for the enormous bridge (4 x $0.41 vs 1 x $2.68). Thermally they're pretty much identical: same theta JA for the collection of four diodes. But the weight and manufacturing cost is less, and Aavid passes the savings along to us, their customers.

So the net delta-money is (4*1.52) + (4*0.41) - (1*1.80) - (1*2.68) which equals $3.24. And what I get for that extra money is: delight. Delight at the knowledge that the rectifiers are avalanche rated. (Bridges aren't.) Delight at the knowledge that the max continuous forward current is 30 amps instead of 25, more safety margin. Delight at the knowledge that these are soft recovery diodes, with the recovery softness parameters specified right on the datasheet, by the manufacturer. Delight at the measured reduction in transformer ringing (this diode and the GBU2510 bridge were both measured in the Linear Audio article and this one's ringing was 17x less than the GBU2510's ringing (V12 @ 2.0A)).

But delight is a subjective thing. That's why economists sweep it under the rug and say that each person has their own, unique, "Utility Function". You may not feel these kinds of delight are worth an extra $3.24. I sure do.
donovas said:
1n400x series seem desirable for me since they're 'soft recovery' types, while being cheap.
Are they? I thought the 1N40xx series are cheap precisely because they are not soft recovery.

Mark Johnson said:
Delight at the knowledge that the rectifiers are avalanche rated.
Why is that relevant to low voltage rectification where diodes are not arranged in series?
I think you may have misunderstood the 1N40xx datasheet. 1A is not the "current limit" but the current rating. The diode does not limit current to 1A. It will pass as much current as you choose to take from it, but if you exceed around 1A for too long then the diode will fail - it will either go short-circuit or open-circuit so in either case will stop being a diode. JMFahey already told you this in post 3.

Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
Why is {selecting an avalanche-rated rectifier} relevant to low voltage rectification where diodes are not arranged in series?
It's just another objective measure of robustness. A positive indicator. Diodes tested and guaranteed with an avalanche rating, have passed more tests of robustness than diodes not-rated. Although with a 4-diode bridge and big filter capacitors and a ~ ±28V DC output, avalanche itself is highly unlikely, as you suggest.
my primary interest is their current limit- 1A. is that still enough to drive 1875 to healthy limit or am i undersupplying the chip?
you have two conditions to check:
a.) the peak repetitive charging current passing to the smoothing capacitors.
b.) the rms current for continuous duty.

I think you will find that the 1A rating of the 1n4004 is too low for both those conditions when driving a power amplifier.

The 3A 1n5404 is big enough.

The peak current demanded by the speaker is fed through the amplifier from the smoothing capacitors.
The rectifier does NOT supply the transient speaker current.


2009-09-09 2:55 am
Sorry for the slightly off-topic but I was wondering the other day what's the point of the lower than 4007 voltage rating diodes, apart from the voltage is there any advantage of using the 4001 for 12v PSU instead of the 4007? I don't think there's a cost difference, last time I checked they were as cheap as a 1/2 watt resistor. I'd be surprised if they are still being produced.

Mark Johnson

Paid Member
2011-05-27 3:27 pm
Silicon Valley
I suppose you could lay out your circuit board to accept axial lead diodes large and small, then try several different kinds of diodes to see whether some of them sound better than others. Here are six through-hole diodes in axial lead packages that I suggest you consider

6A6: 6 amps, 600 volts
1N5404: 3 amps, 400 volts
UF5404: 3 amps, 400 volts, ultra fast
MUR460: 4 amps, 600 volts, fast recovery
VSB3200: 3 amps, 200 volts, Schottky
SR5200: 5 amps, 200 volts, Schottky

All of them are in stock at regular electronics distributors, and you can buy four of each type of diode, for a total cost less than twenty dollars. You might discover that you really enjoy listening to different rectifier diodes in audio equipment. You might also discover that the diodes YOU like, are not the diodes other listeners prefer.