Bias adjustment = better bass if it was low? Allways ?

Excellent! Explained very well and thank you! So at the minimum 14mv bias setting the amp is running 100% in class B all the time?

At 24mv (the manufacture’s highest recommended bias) the amp is running at maximum class A possible within the amp’s design and heat sink capability etc?

That depends on the value of the emitter resistors. The calculation is as follows:
The quiescent current is found by dividing the voltage drop across the emitter resistor by the resistance of that resistor, so:
10mV (0.01V) drop across a 0.22ohm resistor will give..............0.01/0.22 = 0.045A (45.45mA)
24mV (0.024V)drop across a 0.1ohm resistor will give...……….0.024/0.1 = 0.024A (24mA)
14mV (0.024V)drop across a 0.1ohm resistor will give...……….0.014/0.1 = 0.014A (14mA)

So you can see that in order to evaluate the test point voltage properly you need to know what value emitter resistors are fitted to the amplifier.
As a guide, 25mA minimum would probably be in order for most output pairs using silicon transistors.
 
So what happens if we take this a little further by increasing the bias current to a higher value?
Well, the first thing that will happen is that the amplifier will produce more heat when in a quiescent state (i.e. switched on with no input signal). Is this a bad thing? The answer has to be a guarded NO - provided that the heat sinking and ventilation is capable of dissipating the additional heat produced. But remember that this will be the minimum heating level. The harder the amplifier has to work, the more heat it will produce, so increasing the bias current AND driving it hard could well cause it to overheat. Now, bearing in mind that the output stage will be working in class A until the current output required exceeds this level, there may be some sonic advantage to raising this quiescent value slightly (providing that one heeds the temperature warnings). It is generally thought that class A amplifiers are better than class AB - maybe, I don't know. But if you want to experiment with this you could double the quiescent current value - use the calculation above - and your amp would stay in class A for a bit longer, probably with little thermal problems (allowing for good ventilation and watching the temperature -no guarantees here!). It depends how brave you are!

It's worth bearing in mind that with most people, using a system at normal listening volume, the system will spend most of it's time outputting a few milliwatts (in class A ?), only outputting watts or tens of watts during musical peaks.
 
I have experimented extensively with this exercise on my various GAS amps, LSR&D Leach amps, various Bedini amps, Altec 9440A, Boothroyd Stuart Meridian 105. I have found that summing the value of all outputs, that is BJTs, to about 150ma per channel is optimum for an AB amp. But the thing that actually results in a benefit is matching one channel to the other exactly, and I mean perfectly. The only way to successfully match them is to first get them as close as possible with your dvm and then pick either left or right and turn that pot while listening to music, matching it to the other. It will 'tune in' like a radio station. A one turn pot is next to impossible. I replace them with Bourns 25 turn pots. Even then it's a very subtle nudge slowly back and forth. You will hear it equalize. When it does, everything, not just bass will be markedly improved. You now are finally actually hearing stereo. Before it was just different sound coming out of each speaker. This is a very delicate balance and without it there is no true stereo.

I have made acrylic covers essentially matching the originals, ventilation wise, to make sure thermal changes don't take place after adjustment. See through covers with holes placed appropriately allow for adjustment in real world conditions.

You will discover the true potential of your gear
 
Last edited:
The only way to successfully match them is to first get them as close as possible with your dvm and then pick either left or right and turn that pot while listening to music, matching it to the other.

Sounds intriguing. Please go over your "matching" process in a bit more detail. What exactly are you adjusting, measuring, and listening to in order to achieve the optimum match? Will this work for multi-channel amps as well as it does for stereo ones?
 
Sounds intriguing. Please go over your "matching" process in a bit more detail. What exactly are you adjusting, measuring, and listening to in order to achieve the optimum match? Will this work for multi-channel amps as well as it does for stereo ones?
I'm not a tech. I would assume that if in a multi-channel amp each channel has it's own outputs/pots, the process would be equally beneficial.

I'm matching quiescent current with the bias pots. After initial dvm, it's all ears, sitting right between the speakers with the amp.

What I'm listening for is dimensionality. All of a sudden, sound sources are placed in space, seemingly not eminating from either speaker. Hard to describe since the usual descriptors are typically used erroneously. Here is where it really happens, or, in a real sense.

Transients are so clear and pure as to ring like a bell.
 
Last edited:
^^ Very interesting stuff! ^^

I had the covers off again today and should have looked at the values of the resistors. I don’t have a schematic for either of these amps :(
 
As far as I have researched, this follows typical design convention. Increasing the number of outputs serves to share the load, thereby increasing reliability by keeping the outputs cooler at a given power output.
ok so should the typical 2 transistors npn pnp share the same 150 milliamps as one with say 4 outputs npn pnp in a parallel configuration ?
 
I'm matching quiescent current with the bias pots. After initial dvm, it's all ears, sitting right between the speakers with the amp.

What I'm listening for is dimensionality. All of a sudden, sound sources are placed in space, seemingly not eminating from either speaker. Hard to describe since the usual descriptors are typically used erroneously. Here is where it really happens, or, in a real sense.

Transients are so clear and pure as to ring like a bell.

Thanks for the info. Let's see if I've got this right (assume stereo amp with per channel adjustable bias pots):

1. Set bias current (voltage) to manufacturer's spec by adjusting both channel's bias pots to specified current (used hereafter) using DVM.

2. Once idle bias current set, run a (stereo/mono?) music source into the amp and take listening position exactly between the speakers.

3. Adjust bias pot on one channel to produce MORE current (because you don't want to go UNDER the manufacturer's spec, right?) until "dimensionality" and clearness of transients ("ring like a bell") is maximized.

(4. For multichannel amps use same base channel as in Step 3., add a new second channel, and adjust the new channel for max. dim. and c.of t. Repeat Step 4. for all other channels, careful to use same base channel throughout.)

Do I understand your "bias adjustment to ear" procedure correctly?
 
ok so should the typical 2 transistors npn pnp share the same 150 milliamps as one with say 4 outputs npn pnp in a parallel configuration ?
I would think so. But I understand not all outputs are rated the same voltage. I have generalized in reference to the amps I've mentioned. Point being that equilibrium is most important.
 
Thanks for the info. Let's see if I've got this right (assume stereo amp with per channel adjustable bias pots):

1. Set bias current (voltage) to manufacturer's spec by adjusting both channel's bias pots to specified current (used hereafter) using DVM.

2. Once idle bias current set, run a (stereo/mono?) music source into the amp and take listening position exactly between the speakers.

3. Adjust bias pot on one channel to produce MORE current (because you don't want to go UNDER the manufacturer's spec, right?) until "dimensionality" and clearness of transients ("ring like a bell") is maximized.

(4. For multichannel amps use same base channel as in Step 3., add a new second channel, and adjust the new channel for max. dim. and c.of t. Repeat Step 4. for all other channels, careful to use same base channel throughout.)

Do I understand your "bias adjustment to ear" procedure correctly?
Yes, but increasing quiescent current can be advantageous also. Stereo is much better in finding the sweet spot . As I mentioned, once you're satisfied you've done your best with your dvm in terms of equalizing both channels(just because your dvm tells you it's at spec doesn't mean it actually is. It will fluctuate within a range anyway). So matching is a matter of tuning one channel to the other regardless what direction achieves it.
 
Yes, but increasing quiescent current can be advantageous also. Stereo is much better in finding the sweet spot . As I mentioned, once you're satisfied you've done your best with your dvm in terms of equalizing both channels(just because your dvm tells you it's at spec doesn't mean it actually is. It will fluctuate within a range anyway). So matching is a matter of tuning one channel to the other regardless what direction achieves it.

Your last sentence doesn't make sense to me.

Assume a 5-channel amp, with all channels initially set to manufacturer's specified bias current using DVM. If you "go the wrong way" (i.e. reduce current) on one of the channels in your first stereo pair for the listen/match test, you'd be setting its idle current to below the manufacturer's spec. If you the did the same for the other channels as you form up new stereo pairs for your listen/match test and adjustment, you could conceivably end up with only a single channel out of the five at the bias spec and all other channels operating somewhat or significantly below the manufacturer's spec. That doesn't sound like a winner to me as many things (all negative) could be affected by output legs starved of the current needed to operate in their linear output range.
 
Your last sentence doesn't make sense to me.

Assume a 5-channel amp, with all channels initially set to manufacturer's specified bias current using DVM. If you "go the wrong way" (i.e. reduce current) on one of the channels in your first stereo pair for the listen/match test, you'd be setting its idle current to below the manufacturer's spec. If you the did the same for the other channels as you form up new stereo pairs for your listen/match test and adjustment, you could conceivably end up with only a single channel out of the five at the bias spec and all other channels operating somewhat or significantly below the manufacturer's spec. That doesn't sound like a winner to me as many things (all negative) could be affected by output legs starved of the current needed to operate in their linear output range.
Yes of course you are correct. Keeping them within their linear operating "range" is obviously the goal. The problem is they all fluctuate within this range. Listening for optimal settings once they're in there is the most precise way to match their behavior.
 
Got it. Thanks for the novel idea. I've got a 7-channel coming off the bench right now. I'll give your method a try and let you know how it goes (and I'll quantify the difference between mfg spec and the settings achieved by your listen/match procedure).
 
Each transistor of an output pair will have an emitter resistor. Your test point where you measure the volt drop will be across one of them (usually the positive one). The bias current setting applies a standing voltage to both output transistors simultaneously. Even though you are only measuring across one of them, the voltage across the other one will be similar (allowing for transistor gain variations and resistor tolerance). The total minimum current being drawn by the output pair will be double that calculated through a single measurement point.
It doesn't matter how many output channels and amplifier has (not A and B outputs, but actual channels), each channel is a single amplifier which will have it's own bias circuit.
The object of the exercise is to prevent the any of the output transistors from falling below it's minimum point of forward conduction. The precise value of this will vary slightly with different transistor types, but as a general rule should always be above about 25mA emitter current. The actual minimum set voltage drop measured to obtain that value will vary according to the value of the emitter resistor (as shown above).
You must only ever set the bias with the amplifier at normal working temperature and with no music input connected - you are setting the output stage quiescent current and you can only do that if the amplifier is not amplifying anything. Once you have a music signal the voltage drop across the emitter resistors will vary in response the current flowing through the output transistor which will be varying according to the music signal it is amplifying. Trying to set the bias on an amplifier which is running a music signal is a) impossible and b) very dodgy.
 
Back
Top Bottom