diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Pass Labs (http://www.diyaudio.com/forums/pass-labs/)
-   -   Voltage effects on DC offset and bias (http://www.diyaudio.com/forums/pass-labs/207460-voltage-effects-dc-offset-bias.html)

albertNL 24th February 2012 12:18 PM

Voltage effects on DC offset and bias
 
I have been wondering about something and hope you have an answer for me.

If I build an amp, let's say an F4, I understand how and why I need to check DC offset and bias. So far no problems.

Let's assume that I have plenty of heatsink and that my caps are rated high enough. What happens when I replace my transformer by one with a higher or lower output voltage?
Will the DC offset still be OK?
I assume that it will be, because I assume that the DC offset is (to put it mathematically) a function of the parts and connections in the circuit (and the rail voltage will play only a minor role). Is this correct?

And what happens to the bias?
What happens when the rail voltage is 30% higher?
Or 30% lower?

I assume that both DC offset and bias are not be affected (much) by rail voltage because otherwise the +/- 10-15% voltage fluctuation in domestic AC supply would be enough to screw up the settings. Confirmation would be nice though :)

AudioSan 24th February 2012 12:49 PM

DC offset can be effected by offset of +/- rails supply. so transformer switch can effect it. anyway. you will allways check DC offset and bias after you have worked on the amp.

albertNL 24th February 2012 02:17 PM

Quote:

Originally Posted by AudioSan (Post 2920564)
DC offset can be effected by offset of +/- rails supply. so transformer switch can effect it. anyway. you will allways check DC offset and bias after you have worked on the amp.

Thanks Audiosan, I understand that I should always check DC offset and bias.

I am trying to understand how a change in rail voltage will affect DC offset and bias.

AudioSan 24th February 2012 02:46 PM

if +rail go up 10% and -rail does not, DC offset goes up on pluss side.
transformer sec is not allways 100% same. so that can effect DC offset. but not by alot.

albertNL 24th February 2012 03:24 PM

Quote:

Originally Posted by AudioSan (Post 2920685)
if +rail go up 10% and -rail does not, DC offset goes up on pluss side.
transformer sec is not allways 100% same. so that can effect DC offset. but not by alot.

Audiosan, I understand what you mean, but I am not so much interested in the practical problems that can occur. Just the theory.

For the sake of argument, let's say the rail voltages are +24.000V and -24.000V and DC offset is 0V and bias is 1A.
What happens to DC offset and bias when rail voltages are changed to +30.000V and -30.000V?
Or to +18.000V and -18.000V?

AudioSan 24th February 2012 03:34 PM

in theory. nothing.

ottopilot 24th February 2012 11:36 PM

I'm no EE, but my take so far is that any change in rail voltage will cause a corresponding change in dissipation for any given bias due to Watt's law. More voltage = more dissipation. More voltage also means stressing the parts more, hence the reason for cascodes. I also surmise that a change in rail voltage will ultimately change the bias as well. DC offset shouldn't change as others suggest, provided both rails increase or decrease by the same amount. Anecdote: My NAD 2200 has two rails -- 65v and 95v, and let me tell you, even when running at idle on the 65v rail with minimum bias (AB), those sinks get quite warm. It's one of those amps you can smell across the room after it's been on an hour or so. Ah, I love the smell of burning amps in the morning. Smells like, victory!

AudioSan 25th February 2012 12:36 AM

Quote:

Originally Posted by ottopilot (Post 2921482)
I'm no EE, but my take so far is that any change in rail voltage will cause a corresponding change in dissipation for any given bias due to Watt's law. More voltage = more dissipation. More voltage also means stressing the parts more, hence the reason for cascodes. I also surmise that a change in rail voltage will ultimately change the bias as well. DC offset shouldn't change as others suggest, provided both rails increase or decrease by the same amount. Anecdote: My NAD 2200 has two rails -- 65v and 95v, and let me tell you, even when running at idle on the 65v rail with minimum bias (AB), those sinks get quite warm. It's one of those amps you can smell across the room after it's been on an hour or so. Ah, I love the smell of burning amps in the morning. Smells like, victory!

we'r not talking about dissipation. but bias. ex. 1.3A at 18V and it still will be 1.3A at 24V.

albertNL 25th February 2012 08:03 AM

I think I understand that theoretically DC offset will not change by raising or lowering rail voltage. I think this is not the case for the bias, though.

In the Son of Zen (http://passdiy.com/pdf/sonofzen.pdf) amp, the bias is a result of the power drawn by resitors R1-R7. I assume that this is a simple application of Ohm's law (V = I x R). If that is indeed the case, this must mean that the bias is directly proportional to the voltage!

Since R is (effectively) constant in this circuit, V/I must be constant and hence, changing the voltage by any percentage must change the current by the same percentage.

Two questions:

1.
Is the above correct (i.e. in a circuit like Son of Zen, bias depends on voltage)?

2.
What happens in other circuits, which use different topologies to create a constant current source?

PRR 26th February 2012 12:53 AM

> I assume that this is a simple application of Ohm's law (V = I x R). If that is indeed the case, this must mean that the bias is directly proportional to the voltage!

You have over-simplified "simple application".

The SoZ is incredibly stable against supply change. Many circuits are not so stable. Some are far-far fussier.

In the SoZ, *neglecting the cross-Source resistor*, the bias current is the resistors and the voltage from V- to Source. Source voltage is the zero volts at Gate plus the MOSFET's gate-source voltage at that current.

So we have Vgs which is not precisely defined, and the Vgs/Is variation which is also not precisely known.

Therefore V or I are not constant. (Yes, V/I is constant: it's a resistor.)

If V- is huge compared to Vgs, we can wave this away. If V- were -300V, then Vgs of 2V or 5V would be 302V or 305V, 1% change for extreme MOSFETs. However in this application the V- is more likely 30V.

In this circuit, if the MOSFETs are from the same lot (same die size and mask and cooking), then if they gave tolerable match and offset at 25V they would *probably* be OK at 35V. Is and VGS changes, but less than V-, and by nearly the same ratio in both devices of the pair.

We can not neglect the cross-Source resistor. If the change in supply and current causes a 0.1V shift in Vgs, one more than the other, that would be 0.1V/1 or 0.1 Amps offset change, and something less than 0.8V offset change at the speaker. This won't melt a speaker, or cone-offset a stiff woofer, but might put a low-excursion high efficiency driver out of its magnetic center.

The bias current will change nearly proportional to the voltage. This is correct. We apply more supply voltage to get more speaker voltage. With more speaker voltage the speaker current must increase. So we want the bias current to increase.

What else happens? A 30% increase of supply voltage is also a 30% increase of supply current, and a 69% increase of HEAT. If it was idling at 300 Watts heat before, now it idles at 507 watts heat. If the heatsinking was just-OK before, 69% more is likely to be a hot time melt-down.

> What happens in other circuits, which use different topologies to create a constant current source?

Other things. Use Ohms Law carefuly, and think about the whole circuit.


All times are GMT. The time now is 08:57 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2