| beppe61 |
I read that increasing the bias (provided that the amp does not get too hot of course) often is beneficial for the sound.
Some effects can be noted like low end tightening up, crossover and low level signals distortion reduction, ecc.
Is this true?
My amp is solid state and cheap but I have notice that it sounds perceptibly better when it reaches a warmer temperature.
The problem is that with no signal applied at the input the amp remains basically cold.
The heatsink is quite large and I think it could dissipate the more heat generated by the output devices for the bias increasing.
I read that most of cheap commercial amps are biased exaggerately low for safer operation and maybe to protect weak components.
Typical idle bias current in a class AB cheap commercial amp can be as low as 6mA.
I would like to increase the bias current just a bit, in order to reach a higher but steady temperature on the heatsink.
I have already read that this practice is not such uncommon.
I wonder what should be a normal value for the bias current.
The output devices are 15A Toshibas, 2 pairs/channel.
Any suggestion would be extremely helpful.
If this practice can provide beneficial effects I am willing to take some risks.
Thank you very much.
Kind regards,
beppe |
|
|
| anatech |
Hi beppe61,| quote: | | I read that increasing the bias (provided that the amp does not get too hot of course) often is beneficial for the sound. |
This is not always true. It depends greatly on the circuit.| quote: | | Typical idle bias current in a class AB cheap commercial amp can be as low as 6mA. |
Actually, typical bias is on the order of 15 ~ 30 mA per device. If you measure 6 mA it may be due to a special circuit requirement, or the bias is wrong. Just plain misadjusted.| quote: | | I read that most of cheap commercial amps are biased exaggerately low for safer operation and maybe to protect weak components. |
Entirely possible, but not always. Sometimes the distortion may increase at higher bias levels.| quote: | | I wonder what should be a normal value for the bias current. |
There are many other factors. What is the make and model? What are your supply voltages?
Do not believe everything you read. If in doubt, follow the manufacturers recommendations.
-Chris |
|
|
| AndrewT |
Hi,
just to confirm, you have 8 Toshiba 15A devices in your stereo amplifier output stage?
What is Iq just now?
What are your Vrail voltages?
Is it an EF driver and output stage?
or
a CFP driver and output stage?
Is it complementary?
or
Quasi?
If you can keep the sink temperature below 35degC then you could try raising Iq but this is not to gain any benefit from increased temperature but to ensure the output stage does not unbias itself and cause low bias induced distortion . |
|
|
| quasi |
Did someone say Quasi?
If you have a schematic of the power amplifier we can probably give better advice. Higher biasing can help reduce distortion especially at lower volumes but there is no point in trying to get it from say 0.01% to 0.005%.
If you double the bias you will double the heat and double the bias will also increase the residual ripple on the power supply (especially on consumer grade amps).
Post a schematic if you can.
Cheers
Quasi |
|
|
| EchoWars |
| quote: | Originally posted by quasi
Did someone say Quasi? |
:D
I see blown amps all the time from guys who thought that 'if a little bias is good, a lot must be better'.| quote: | | If in doubt, follow the manufacturers recommendations. |
If you like listening to music rather than replacing blown transistors and burnt resistors, this is the mantra to repeat to yourself. |
|
|
| Leolabs |
| Hi beppe61.Take a trip to www.tnt-audio.com on their DIY article regarding increasing bias of the amp. |
|
|
| sam9 |
I seen that article. IMO one of the more irresponsible audio posts I've seen. As was mentioned above the details of the output stage more than a little important. As I recall the the article addresses this little or not at all.
I suspect that there may be a basis for the article in dealing with MOSFET output devices. I'm not sure but I think what happens is increasing the bias moves them closer to class A and move the cross over notches (or "half-notches') away from the zero crossing point.
My personal experience trying to set optimal bias on EF output sections is that: A- "optimal" is elusive and requires a lot of fiddling with thermal tracking schemes, B-tyhere is quite a bit more latitude than a lot books and articles would lead you to believe, C- a practical solution is get approximately close to optimal (assuming that's even determinable) and then reduce the bias as much as you can without bringing on audible deterioration -- cooler is better so long as it doesn't adversely affect what you hear. |
|
|
| EchoWars |
| That tnt-audio article is responsible for more dead amplifiers than I care to count. |
|
|
| quasi |
| quote: | Originally posted by EchoWars
:D
I see blown amps all the time from guys who thought that 'if a little bias is good, a lot must be better'. If you like listening to music rather than replacing blown transistors and burnt resistors, this is the mantra to repeat to yourself. |
I have fiddled with bias before without any problems. But (and I mean but), with circuits and applications that I understand. So from my part if you truly know what you are doing and understand completely what you're doing it to, then by all means play. By the way I have sometimes turned the bias down a bit ...again according to application.
Cheers |
|
|
| johnny1 |
Have in mind that manufacturers go for low bias currents because they don't know the ventilation in your rack, or if you drive small bokself speakers or Maggies!
That is the main reason for the low bias current.
My rules of thumb when messing with quiescent current are the following:
1. Don't increase the current with steps higher than 5-10mA
2. Always check the temperature of the heat sinks
3. Heat sinks for B or AB class amps shoud never exceed 50C (empirical rule, based on thermal stability and sound quality observations).
4. An amp designed for 15mA QC will never work reliably with 120mA QC
5. follow rules 1,2,3 and 4!:D
I was in the electronic service bussiness and have seen many blown amps and fried woofers caused by silly tweaks.
Your BJT's are probably the 2SC5200/2SA1943 or the
2SC3281/2SA1302.
Please post a schematic or make/type of the amp
What QC do you have now? |
|
|
| AndrewT |
Hi,| quote: | | main reason for the low bias current |
I suspect the main reason for low bias current is to achieve low production cost. Even to the detriment of sound quality.
BUT, it does not follow that increasing the bias current will increase the sound quality.
If designed properly the amp should leave the assembly line adjusted to best Iq for that set of design compromises. |
|
|
| jez |
I have had several phone calls from people wanting to increase the bias in their amps just lately. TNT to blame eh:whazzat:
I'm sick of trying to explain to numptys that they should leave well alone unless they know as much about the design as the manufacturer does!....it's like that for a reason:o:rolleyes:
The last one to phone me about this wouldn't listen.....he rang back a few days later to say that if it hadn't been for the thermal cutout....you can guess the rest!
A good start would be to read Doug Self's book on this subject= wise words;) |
|
|
| johnny1 |
| quote: | Originally posted by AndrewT
BUT, it does not follow that increasing the bias current will increase the sound quality.
|
Well, that depends on the design.
For example, the toshiba BJT's mentioned above have linear hfe curve for Iq>100mA.
If an amp with those toshibas is running at 15mA, it will benefit from the Iq increase.
edit:--> if the rest of circuit + heat sink can support that!
This is the reason that made Onsemi to introduce the MJL 0281/MJL 0302 which are a great choice for low Iq and low power consumer amps. |
|
|
| beppe61 |
I thank you all Sirs for the extremely kind and valuable reply.
I understand that increasing bias is a very risky move and moreover it is not a guarantee of a better sound (!).
The amp in question is a Behringer A500.
I asked Behringer for a service manual and I got a no.
But what I wanted to know above all other things is if there is a sure correlation between bias level and quality of sound.
Now I know that high bias does not guarantee high quality sound.
It depends on the actual design.
This is very fundamental to me.
Now I wonder if there are, on the contrary, low bias high sound quality amps.
Thank you very much indeed.
Kind regards,
beppe |
|
|
| BlackCatSound |
Doug Self's amp design book goes into some detail about ideal bias on an A/B amp.
High bias just shifts the distortion points away from the zero crossing point. Effectively giving you two places where distortion occurs rather than just 1.
Doug's Class XD does reduce this effect. I did have a link to the tech details of class XD but I've lost it. Sorry!
There is some seriously questionable advice on TNT audio, oh well, there will always be sites like that. |
|
|
| AndrewT |
Hi,| quote: | | low bias high sound quality amps |
are unlikely to be FET output stage.
Much more likely to be BJT.
Crimson have done a low bias quasi for decades that sounds pretty good using one pair of 2n3773.
They have moved on to complementary recently but I have no info on bias but suspect it is also low bias (just to maintain reliability with the plastic packaged devices). |
|
|
| janneman |
| quote: | Originally posted by johnny1
[snip]My rules of thumb when messing with quiescent current are the following:[snip] |
Don't mess with QC, your chances of going backwards are 10 to 1.
Jan Didden |
|
|
| janneman |
| quote: | Originally posted by johnny1
[snip]For example, the toshiba BJT's mentioned above have linear hfe curve for Iq>100mA.
If an amp with those toshibas is running at 15mA, it will benefit from the Iq increase.[snip |
Not true. The final sound quality with different bias currents has NOTHING whatsoever to do with the Ft at that particular bias point.
The bias point is just that, a bias point. From there on, the amp starts to work with varying currents and voltages and frequencies. The Ft, and its variations, while working with all those different frequencies and currents and voltages, have an impact on open loop gain versus frequency, and that, together with the feedback factor, *may* influence the sound.
Jan Didden |
|
|
| johnny1 |
Jan,
I didn't mention Ft, i mentioned hfe(current gain) |
|
|
| beppe61 |
| quote: | Originally posted by johnny1
Well, that depends on the design.
...
|
Very interesting and kind reply Sir. Thank you.
Let me quote some your sentences.
For example, the toshiba BJT's mentioned above have linear hfe curve for Iq>100mA.
If an amp with those toshibas is running at 15mA, it will benefit from the Iq increase.
If I understand well given a certain device it has an optimal working point where it is most linear, for instance.
So more than on design it depends on the specific device.
Am I wrong?
The Toshi bjt mentioned working at 15 mA it just does not work optimally and the design that makes the Toshi working like that is not optimal.
edit:--> if the rest of circuit + heat sink can support that!
I agree completely on this.
Actually I understand the importance of a large heatsink surface.
This is the reason that made Onsemi to introduce the MJL 0281/MJL 0302 which are a great choice for low Iq and low power consumer amps.
That is very interesting.
So low bias (i.e. consumption) power amps are possible.
Thank you very much again. Interesting topic this for me.
Kind regards,
beppe |
|
|
| janneman |
| quote: | Originally posted by johnny1
Jan,
I didn't mention Ft, i mentioned hfe(current gain) |
Indeed, I stand corrected. I was *thinking* of Hfe by wrote Ft.
But I still maintain my general reasoning.
One issue not mentioned for instance is gain doubling at the bias point because both transistors contribute to the gain, whereas in the class B area only one contributes. This is shown in the gain versus output voltage curve where the gain flattens to half outside the class A area. So, biasing in an area where Hfe is half the value of the higher current region may actually be quite beneficial as it compensates for the gain doubling.
Jan Didden |
|
|
| MikeB |
There is an "optimal" bias for ClassAB EFs (using BJTs), the point where GM-doubling and unlinearity of vbe for low currents cancel out each other. (not 100%, but nearly)
This value is ~24/25mv drop over the emitter resistor. Above and below this bias the distortion rises. Of course pure ClassA has the lowest distortion. But at some point the ClassA operation is left, and the ClassB operation takes over, and these distortions are there again.
This gives ~109ma each pair using 0.22ohm emitter resistors. (or 53ma using 0.47ohm for 25mv, greetings to Hugh ! ;) )
On the other hand, higher bias leaves more time for the ouput devices to open/close. But modern high speed bjts like the MJL0281/0302 are fast enough...
Mike
EDIT: Jan was 1 minute faster... :D |
|
|
| johnny1 |
bepee,
The behringer A-500 is a pro amp without fan.
The designer knows that it will probably work 24h/day, 7days/week etc...
What he does not know is how the customer will use this pro gear.
And pro gear should always work, and never fail.
So, don't say that the design is not optimal.
It is OK for what it is intended to do.
Personally, if the amp was mine, (for home use) i would probably make some tweaks on it.
BUT
You need the schematic, you should know what you are doing, and no one in this site can guarantee that everything will be OK.
So, in other words you are on your own;) |
|
|
| beppe61 |
| quote: | Originally posted by MikeB
There is an "optimal" bias for ClassAB EFs (using BJTs), the point where GM-doubling and unlinearity of vbe for low currents cancel out each other. (not 100%, but nearly)
This value is ~24/25mv drop over the emitter resistor. Above and below this bias the distortion rises. Of course pure ClassA has the lowest distortion.
But at some point the ClassA operation is left, and the ClassB operation takes over, and these distortions are there again.
This gives ~109ma each pair using 0.22ohm emitter resistors. (or 53ma using 0.47ohm for 25mv, greetings to Hugh ! ;) )
On the other hand, higher bias leaves more time for the ouput devices to open/close.
But modern high speed bjts like the MJL0281/0302 are fast enough...
Mike
EDIT: Jan was 1 minute faster... :D |
Thank you Mr. Mike. Very interesting.
Regards,
beppe |
|
|
| janneman |
| quote: | Originally posted by MikeB
[snip]EDIT: Jan was 1 minute faster... :D |
... but I forgot to mention that optimum Vre..... "Haastige spoed is zelden goed" as we say in NLD.
Jan Didden |
|
|
| beppe61 |
| quote: | Originally posted by johnny1
bepee,
The behringer A-500 is a pro amp without fan.
The designer knows that it will probably work 24h/day, 7days/week etc...
What he does not know is how the customer will use this pro gear.
And pro gear should always work, and never fail.
So, don't say that the design is not optimal.
It is OK for what it is intended to do.
Personally, if the amp was mine, (for home use) i would probably make some tweaks on it.
BUT
You need the schematic, you should know what you are doing, and no one in this site can guarantee that everything will be OK.
So, in other words you are on your own;) |
Thank you very much Mr. Johnny1 for your kind and sensible advice.
All my ramblings start from the feeling that the amp sounds better when warm.
At idle without signal the amp stays cold.
Only after some playing music on it the sounds get better.
In my silly mind, a slight increase in the bias current should give a increase in the amp temperature and consequently a better sound.
I understand perfectly well that the temperature of the amp must be kept under control during this "experiment".
My idea would be to try with a new bias current = 20mA that from what I have got here is a very common value for commercial amps.
Thank you sincerely for preventing me from some "insane" action.
Kind regards,
beppe |
|
|
| BlackCatSound |
| the obvious thing to do is take distortion measurements for different bias currents. |
|
|
| richie00boy |
Beppe,
Have you tried letting the amp play into a dummy load then connecting your speakers for a listen and seeing if the sound is better? My reasoning is that you think it sounds better when music has been playing a while, is due to your ears getting used to the sound in the 'warm up' period.
I must add that I have experienced this phenomenon of sounding better after a time, but I think changing bias current is not anything to do with it. |
|
|
| beppe61 |
| quote: | Originally posted by richie00boy
Beppe,
Have you tried letting the amp play into a dummy load then connecting your speakers for a listen and seeing if the sound is better? My reasoning is that you think it sounds better when music has been playing a while, is due to your ears getting used to the sound in the 'warm up' period.
I must add that I have experienced this phenomenon of sounding better after a time, but I think changing bias current is not anything to do with it. |
Thank you for your interesting suggestion.
Actually I have two 5ohm/200W resistors packs to use as dummy loads.
I will try what you suggest as soon as possible.
But let me be a little unscientific.
I have gathered a lot of favourable response about the nice effects consequent to bias current increase that this practice intrigues me more and more.
I understand it could be risky.
Nevertheless if the bias current is extremely low I understand that it is a very uncommon situation.
Regards,
beppe |
|
|
| richie00boy |
Basically I'm saying that the bias current and the need to warm up are two totally different things.
The bias current does not usually change drastically when going from cold to warmed up, so I do not believe that you changing bias current is the way to achieve what you seek.
However I do appreciate that in addition to the warm up issue, the bias current could also possibly be adjusted to optimise the sound. |
|
|
| beppe61 |
| quote: | Originally posted by richie00boy
Basically I'm saying that the bias current and the need to warm up are two totally different things.
The bias current does not usually change drastically when going from cold to warmed up, so I do not believe that you changing bias current is the way to achieve what you seek.
However I do appreciate that in addition to the warm up issue, the bias current could also possibly be adjusted to optimise the sound. |
I understand your points and I thank you for the advice.
I will try the procedure you mentioned before going on with any "adjustment".
Regards,
beppe |
|
|
| richie00boy |
| That's a good approach Beppe, look forward to reading your findings. |
|
|
| pooge |
| quote: | Originally posted by janneman
One issue not mentioned for instance is gain doubling at the bias point because both transistors contribute to the gain, whereas in the class B area only one contributes. This is shown in the gain versus output voltage curve where the gain flattens to half outside the class A area. So, biasing in an area where Hfe is half the value of the higher current region may actually be quite beneficial as it compensates for the gain doubling. |
I'm not saying you are right or wrong, but there is some controversy over this. For example, Self swears by this phenominon, but Leach says it doesn't exist. I don't know who is correct. I tried to engage each in a discussion about it, to no avail. Self said he read Leach's arguments, but said he hadn't analyzed it fully, and would have to study it some more. There may be a difference in semantics clouding the disagreement; but if so, I couldn't get them to resolve it. Anyhow, Leach's arguments are posted here, following Figure 15:
http://users.ece.gatech.edu/~mleach/lowtim/output.html
And here:
http://users.ece.gatech.edu/~mleach/papers/classab.pdf
Don't flame me. I'm just the messenger. I'd like to know the answer to this. |
|
|
| beppe61 |
| quote: | Originally posted by richie00boy
That's a good approach Beppe, look forward to reading your findings. |
I will report of course.
Nevertheless after your question the more I think about it the more I cannot understand why a warm-up amp should sound sound better than cold.
It seems very difficult to explain.
Expecially in considering that the bias current stays the same.
Anyway the answer is in the test.
Thanks and regards,
beppe |
|
|
| sreten |
| quote: | Originally posted by janneman
Don't mess with QC, your chances of going backwards are 10 to 1.
Jan Didden |
Hi, 10 to 1 on, or going forwards is 10 to 1, :)/sreten. |
|
|
| BlackCatSound |
Self goes into gm doubling in some detail with a fair few diagrams and measurements in his book.
Although his output stage is quite different to Leach. |
|
|
| sreten |
| quote: | Originally posted by johnny1
Well, that depends on the design.
For example, the toshiba BJT's mentioned above have linear hfe curve for Iq>100mA.
If an amp with those toshibas is running at 15mA, it will benefit from the Iq increase.
edit:--> if the rest of circuit + heat sink can support that!
This is the reason that made Onsemi to introduce the MJL 0281/MJL 0302 which are a great choice for low Iq and low power consumer amps. |
| quote: | Originally posted by janneman
Not true. The final sound quality with different bias currents has NOTHING whatsoever to do with the Ft at that particular bias point.
Jan Didden |
Hi,
Ft or hfe it doesn't matter, the first post is drivel.
Correct biasing is not related to Iq (Iq is a consequence of biasing
and very dependent on the output stage details, not the devices)
and there is no optimum bias current for particular BJT devices.
There is no such thing as an output transistor specifically developed
to be used with low Iq, you simply don't understand the problem.
(May be developed specifically for audio output stages......)
:)/sreten. |
|
|
| pooge |
| quote: | Originally posted by BlackCatSound
Self goes into gm doubling in some detail with a fair few diagrams and measurements in his book.
Although his output stage is quite different to Leach. |
Also, I think there is some difference in what each considers "Class AB". |
|
|
| anatech |
Hi All,
The fact remains that we are dealing with BJTs that do not require a high bias point, and that each circuit is different. I have seen amplifiers where the total THD does not change from a couple mA to 20 mA (except there is more supply noise). I have also seen power amps that show a steady improvement as the bias current is increased. So, it depends on the design.
In each case, regardless of bias level, there is normally an improvement in all parameters that may take 30 sec to 1/2 hour to settle in. Remember, this is independant of bias level. This with a distortion analyser hung on the output across an 8 ohm dummy load. I agree with BlackCatSound here too. Measure the differences as your bias is increased in steps. We humans are terrible at judging sound when it's our work we are judging. ;)
-Chris |
|
|
| sreten |
Hi,
Class aB is attempted optimal biasing for low distortion.
Class AB is deliberately designed to operate partly in class A.
At powers where both the above operate in class B for similar circuits
the latter always has more distortion. But the the latter will have less
distortion over the range it remains in class A.
:)/sreten. |
|
|
| beppe61 |
| quote: | Originally posted by sreten
Hi,
Ft or hfe it doesn't matter, the first post is drivel.
Correct biasing is not related to Iq (Iq is a consequence of biasing and very dependent on the output stage details, not the devices) and there is no optimum bias current for particular BJT devices.
There is no such thing as an output transistor specifically developed
to be used with low Iq, you simply don't understand the problem.
(May be developed specifically for audio output stages......)
:)/sreten. |
there is no optimum bias current for particular BJT devices.
Thank you Sir. This is important for me to understand.
Is the performance of a BJT in some way related to the bias current?
Is there an optimum bias for a complete amp (not the single device)?
I would like to understand better that issue.
I think is very fundamental.
Thank you very much indeed.
Kind regards,
beppe |
|
|
| MikeB |
| quote: | Originally posted by pooge
I'm not saying you are right or wrong, but there is some controversy over this. For example, Self swears by this phenominon, but Leach says it doesn't exist. I don't know who is correct. |
In this case i'd say Self is correct, i experimentally proved the 24mv in sims, studying the unlinearities in the outputstage... (GM-doubling)
I second sreten, Class-aB has the overall lowest distortion, Class-AB has the lower distortion only at low power levels.
But, this is only valid if the output BJTs do not suffer from crossover distortion at high freqs at optimal bias due to beeing too slow.
Mike |
|
|
| sreten |
| quote: | Originally posted by beppe61
I will report of course.
Nevertheless after your question the more I think about it the
more I cannot understand why a warm-up amp should sound
sound better than cold.
It seems very difficult to explain.
Expecially in considering that the bias current stays the same.
Anyway the answer is in the test.
Thanks and regards,
beppe |
Hi,
It does not stay the same. Even if it did the bias conditions will be
different when an amplifier is warm compared to when it is cold.
Accurate aB biasing is about the point the two halfs hand over
output current on the devices Vbe curves, and Vbe is very
temperature dependent , hence accurate biasing requires
knowledge of the output devices temperature, Iq impacts
Vbe via the drop across the emitter resistors value, Ve.
The bias network produces a voltage which sums to 2 x Ve +
an even number x Vbe depending on output stage topology.
Accurate aB biasing is about the point the two halfs hand over
output current on the devices Vbe curves, and Vbe is very
temperature dependent , hence accurate biasing requires
knowledge of the output devices temperature and
compensating for the changes in biasing conditions.
An amplifer will be adjusted for optimum bias when warm. When
cold it will have non-optimum bias. If the temperature compensation
of the amplifer is poor its bias will be further out when cold compared
to an amplifier with good temperature characteristics.
Doubling bias current is not a good idea.
:)/sreten. |
|
|
| sreten |
Hi,
regarding Self versus Leach :
Self decribes a phenomena he calls "gm doubling", it is very real.
Leach describes a fallacious situation that he says leads to the
concept of "transconductance doubling" which does not exist.
They are not talking about the same thing, Self is talking about overbiasing
whilst Leach is talking about misconceptions of optimum biasing.
They both agree optimum biasing gives the flattest transconductance.
Leach then goes on to ignore the overbiasing case and in
this sense is wrong, because this does illustrate a real case
of "transconductance doubling".
Self has a much better grasp of all the issue involved.
:)/sreten. |
|
|
| ilimzn |
| It should be noted that some high Ft low beta droop BJTs are actually of cell based design, in essence a large number of smaller transistors in parallel, with an emitter resistor integrated in the structure, for each cell (LAPT structure). This is frequently disregarded, even when seen in the datasheets - in form of Vbe ~~ 1.5-2V at maximum Ic, and correspondingly higher saturation voltage. The built-in degeneration should actually be included when exploring optimum bias. |
|
|
| richie00boy |
| quote: | Originally posted by beppe61
I will report of course.
Nevertheless after your question the more I think about it the more I cannot understand why a warm-up amp should sound sound better than cold.
It seems very difficult to explain.
Expecially in considering that the bias current stays the same.
Anyway the answer is in the test.
Thanks and regards,
beppe |
As sreten has pointed out it does not stay the same, but it usually does not vary a lot, unlike what I had presumed you would be doing if you fiddled with the bias current.
As well as the bias current, there is also the issue of semiconductors behaving slightly different when hot.
Continue with your planned test :) |
|
|
| beppe61 |
| quote: | Originally posted by sreten
Hi, It does not stay the same.
Even if it did the bias conditions will be different when an amplifier is warm compared to when it is cold.
Accurate aB biasing is about the point the two halfs hand over
output current on the devices Vbe curves, and Vbe is very
temperature dependent , hence accurate biasing requires
knowledge of the output devices temperature, Iq impacts
Vbe via the drop across the emitter resistors value, Ve.
The bias network produces a voltage which sums to 2 x Ve +
an even number x Vbe depending on output stage topology.
Accurate aB biasing is about the point the two halfs hand over
output current on the devices Vbe curves, and Vbe is very
temperature dependent , hence accurate biasing requires
knowledge of the output devices temperature and
compensating for the changes in biasing conditions.
An amplifer will be adjusted for optimum bias when warm.
When cold it will have non-optimum bias.
If the temperature compensation of the amplifer is poor its bias will be further out when cold compared to an amplifier with good temperature characteristics.
Doubling bias current is not a good idea.
:)/sreten. |
Thank you very much Sir for your valuable and extremely thorough explanation.
I have a long way to go before I can understand all these things.
As I understand the full procedure of bias adjusting is time comsuming my feeling is that when the equipment is so cheap like in my case the manufacturer does even not bother with it.
I will try to actually check it as a first action.
Only in the case it were extremely low and outside of what I understand is the quite normal range (i.e. 15-30mA) I will proceed.
Thank you very much and kind regards,
beppe |
|
|
| beppe61 |
| quote: | Originally posted by sreten
Hi,
regarding Self versus Leach :
...
They both agree optimum biasing gives the flattest transconductance
...
:)/sreten. |
So a optimum bias condition does exist.
This is fundamental for me.
Thank you very much for this information.
However I still wonder if this statement refers to the single device (bjt) or the amp on the whole.
Anyway I think it is a very fundamental statement.
Thank you very very much.
Regards,
beppe |
|
|
| MikeB |
This is about outputstage only. See my posting:
http://www.diyaudio.com/forums/show...066#post1006066
I already gave you the information about "optimal" bias 2 pages ago...
You have to set bias for each stage seperately, depending on topology and targeted power output. These ClassAB informations do not apply to the rest of the amp, none of the devices there should ever leave ClassA in normal operation.
Mike |
|
|
| beppe61 |
| quote: | Originally posted by MikeB
This is about outputstage only.
See my posting: [url]http://www.diyaudio.com/forum...066#post1006066[/url]
I already gave you the information about "optimal" bias 2 pages ago... You have to set bias for each stage seperately, depending on topology and targeted power output.
These ClassAB informations do not apply to the rest of the amp, none of the devices there should ever leave ClassA in normal operation. Mike |
Please excuse me Mr. Mike.
Now I realize better your advice.
Please correct me if I understand badly.
An optimum bias current for all bjts working in Class AB is 109 mA (i.e. 3-4 times the values chosen by manufacturers for the majority of commercial amps).
So the majority of commercial amps having around 15-30mA are misbiased.
Is this correct ?
If so how this "misbiasing" translates in terms of sound?
I have a further question.
I suspect that 100mAs of bias current generate quite a bit of heath and large heatsinks are mandatory.
Unfortunately this is not my case.
I think I will be forced to opt for a lower current anyway, even if this is not the optimum value.
Thank you very much.
Kind regards,
beppe |
|
|
| MikeB |
No problem Beppe. The 109ma is only valid for 0.22ohms. It's about dropping 25mv across the REs. It depends on the REs used in this amp. If they are 0.33ohm the bias would be ~73ma.
But, as Ilimzn pointed out, the LATP transistors might have built in REs...
Mike |
|
|
| jacco vermeulen |
FYI:
LAPT, Linear Amplification Planar Transistor
i've tried the bias tweaking stuff too with consumer stuff, at the time when MF made models with higher bias setting.
imo, for AB stages you are better off with the 25 to 35mV rule.
If you go higher, go much higher or full class A.
Either, Or, not something inbetween. Unless you change the number of parallel devices and/or are able to change emitter resistor values.
Tweak a 100 watts class AB amplifier, go the Marantz way.(PM78-PM94-PM7200) |
|
|
| beppe61 |
| quote: | Originally posted by MikeB
No problem Beppe. The 109ma is only valid for 0.22ohms.
It's about dropping 25mv across the REs.
It depends on the REs used in this amp.
If they are 0.33ohm the bias would be ~73ma.
But, as Ilimzn pointed out, the LATP transistors might have built in REs... Mike |
All right Mr. Mike.
So I have to act on the bias trimmer in order to get 25mV across RE. I could always try while monitoring carefully the temperature on the heatsinks.
What are LATP transistors ?
Anyway I do not think is my case.
I am attaching a view of the interiors when the REs can be seen.
Thank you very much again.
Kind regards,
beppe |
|
|
| MikeB |
Jacco, my fault...
Beppe, regarding the size of the heatsink and the "complexity" of your amp... Don't touch it... Did you see any trimpot for bias ?
Beppe, the REs mentioned by Ilimnz are if the transistor is made of several smaller transistors, in this case these are needed to share current among these.
Mike |
|
|
| jacco vermeulen |
The Sankens we talked about so much are planar epitaxial devices, Beppe.
RETs, ringemitters, multi-emitter transistors.(Sanken, Toshiba, Fujitsu, and the rare ones Pioneer used)
My fault: Planar should be Power. L.A.Power.Transistor
[it's Friday, YeeHaa(@Nicholas Cage Copyright)] |
|
|
| beppe61 |
| quote: | Originally posted by jacco vermeulen
FYI:
(1) LAPT, Linear Amplification Planar Transistor
i've tried the bias tweaking stuff too with consumer stuff, at the time when MF made models with higher bias setting.
(2) imo, for AB stages you are better off with the 25 to 35mV rule.
(3) If you go higher, go much higher or full class A.
Either, Or, not something inbetween.
Unless you change the number of parallel devices and/or are able to change emitter resistor values.
(4) Tweak a 100 watts class AB amplifier, go the Marantz way(PM78-PM94-PM7200)
|
Thank you sincerely Mr. Vermeulen for your kind advice.
1 - thank you for the info. Never seen them.
2 - so another vote for the 25mV value. Do you think that the temperature could increase dangerously?
I will monitor it anyway.
3 - I record the advice.
4 - do you mean that the mentioned models are biased at about 25mV across the REs and so higher than other commercial amps?
Interesting because the PM94 in particular was one of the Marantz most prestigiuos amp if I remember well.
Thank you a lot.
Regards,
beppe |
|
|
| beppe61 |
| quote: | Originally posted by MikeB
Jacco, my fault...
Beppe, regarding the size of the heatsink and the "complexity" of your amp... Don't touch it...
Did you see any trimpot for bias ?
Beppe, the REs mentioned by Ilimnz are if the transistor is made of several smaller transistors, in this case these are needed to share current among these.
Mike |
Actually looking inside I can see 3 pots.
I have to disassemble the amp to expose them.
I am so resolute to try the up-biasing that I could have my amp serviced by a lab I trust completely.
I am pretty sure the present bias is very very low.
I can leave the amp on 24/24 and the heatsink remain completely cold, like the sound actually.
If I had the actual schematic, that is unobtainable, I would try myself.
beppe |
|
|
| EWorkshop1708 |
If you have a high bias, the amp sounds fine when cold.
In the case of larger amps, they take a long time to get hot anyway.
My Sub amp has 46mV across 0.15 ohm emitter resistors, 306mA when fully warm, about 280mA cold. However, I have large heatsinks and low RPM fan cooling. Heatsinks barely warm to the touch. Because of the bias, bass notes at low volume sound tight and controlled.
Either way sounds fine. I see no problem with increasing the bias on an amp as long as the transistors, and heatsink can handle the heat output of the amp.
I also am a fan of lowering bias on an amp that has it set high, or if the amp has a small heatsink. Some amps just run way too hot at idle IMO.
A hot running Sony amp usually runs much cooler with a small decrease in bias.
Just play with the bias, if the amp sounds good with high bias and doesn't get too hot, then go with it. However, if you don't hear much difference, and/or the amp gets too hot, then use a low bias.
It depends on the amp and the heatsinking.
Have fun |
|
|
| sam9 |
Two observations:
1- Regarding the observation that an amplifier may sound better after ~30 of warm up. Certainly plausible. The bias was probably set once a stable thermal cvondition was reached, i.e., after ~30 minutes. Presumably this wouldn't occurr if there were perfect thermal tracking or even "good enough". I can't say I've ever heard a change that I would notice by listening although I've seen a fft spectrum improve a little.
2- Regarding the benefits of AB (the partial Class A definition). I'll buy Self's position that the point of A-to-B transition introduces additional distortion. At the same time, in actual use many people listen at moderate levels and to program sources that may keep the amp[lifier in the Class A range nearly all the time. For them, this may be a good configuration. Maybe a 20W true class A amp would be better still. Decades ago a 20W NAD reciever seemed to me to be more than adequate. |
|
|
| anatech |
Hi sam9,
I have seen an amplifier sound a little better after 5 min, and another after 20. In both cases the bias currents were perfectly stable. I think this has to do more with the front end stages than the output stages.
Then again I've seen more than my share of amplifiers that have horribly unstable bias current and do take a good half hour to become listenable. The early Bryston 3 and 4 B come to mind here as perfect examples of this.
So don't focus on bias current as the cure all for fidelity. There are several other parts of an amplifier that impact this.
-Chris |
|
|
| beppe61 |
| quote: | Originally posted by anatech
Hi sam9,
I have seen an amplifier sound a little better after 5 min, and another after 20. In both cases the bias currents were perfectly stable.
I think this has to do more with the front end stages than the output stages.
Then again I've seen more than my share of amplifiers that have horribly unstable bias current and do take a good half hour to become listenable.
The early Bryston 3 and 4 B come to mind here as perfect examples of this.
So don't focus on bias current as the cure all for fidelity.
There are several other parts of an amplifier that impact this.
-Chris |
Thank you Mr. Chris.
So, as you very clearly state, the bias should not be seen as the cure of all problems.
Other things, like for instance an undersized power supply, are much more detrimental to sound.
Taken the "25mV across RE" rule as a reference for setting the bias then it would be better to adress other parts of the unit for upgrading.
I will have the bias checked in that sense.
Thank you very much and kind regards,
beppe |
|
|
| sreten |
| quote: | Originally posted by beppe61
Taken the "25mV across RE" rule as a reference for setting the bias
beppe |
This is only applicable to a particular output configuration,
if the actual bias current is 10mA it is highly unlikely that this
rule is applicable as its likely the output configuration is some
sort of of CFP stage, which has much lower standing current.
:)/sreten. |
|
|
| beppe61 |
| quote: | Originally posted by sreten
This is only applicable to a particular output configuration
if the actual bias current is 10mA it is highly unlikely that this
rule is applicable as its likely the output configuration is some
sort of of CFP stage, which has much lower standing current.
:)/sreten. |
So to get a schematic is mandatory before proceeding with any action.
I think that it should be very similar to the Alesis RA 300.
Actually they seems twins.
regards,
beppe |
|
|
| jacco vermeulen |
| quote: | Originally posted by anatech
this has to do more with the front end stages than the output stages. |
Build stuff with separate PS's for front end and driver/output stage.
Switch off the power for the output stage but keep the front end alive.
Adds only a few pennies to the electricity bill, but heat-up time is dramatically reduced. Even on big Class A amps, additionally keeping the output stage alive but switched to a low bias setting and you''ll have 99% performance in a few minutes. |
|
|
| AndrewT |
Hi,| quote: | | keeping the output stage alive but switched to a low bias setting and you''ll have 99% performance in a few minutes. |
I knew the Krell Klone had intended the low bias switch to be for a standby relay. |
|
|
| sam9 |
This bias business, at least as presented in the TNT article is sympyomatic of a certain tendancy thought in audio (and probably other areas of human endevor as well). Namely, if a little more (or less) of something is helpful in a specific instance still more will will aways be helpful regardless of circumstance.
If one particular manufacturer uses far too little filter capacitance in the power supply then it is reasoned that all amplifiers can be improved by increaseing the size of filter caps without limit.
If one amplifier comes from the factory with bias set too low then it must be true that all amps everywhere will benefit from increasing the bias.
If someone finds speakers hooked up with 50 feet of 28AWG and finds it helpful to replace it with16AWG, then it follows that a 6 foot run of 12GA should be replaced by 6GA
ECT.
At least thi approach is easier than learning and thinking. |
|
|
| jacco vermeulen |
I'm sorry,
i was busy sniffing a pound of Afghan opium, what were you saying ? |
|
|
| janneman |
| quote: | Originally posted by sreten
This is only applicable to a particular output configuration,
if the actual bias current is 10mA it is highly unlikely that this
rule is applicable as its likely the output configuration is some
sort of of CFP stage, which has much lower standing current.
:)/sreten. |
Doug Self makes a strong case that it IS applicable to ANY (bjt) output config, be it EF or CFP. It has to do with the relation between the voltage across Re and Vbe.
There is of course a difference in Ibias for the different output stage configuration when the Vre is the same. The EF stage will have almost an order of magnitude larger Iq than the CFP for optimal Vre bias. Your quoted 10mA is Dougs recommendation for Re=0.22 ohms (actually, 11.5mA at Vbias=1.297V).
Jan Didden |
|
|
| p robertson |
| i have a p3a amp,and re=0.33ohms,what does DSelf say is the optimum bias in this case?thanks |
|
|
| jaycee |
Better to ask Rod Elliot that question :)
Most people seem to bias their P3A's for 30mA quiescent current though. |
|
|
| beppe61 |
Just to state again an important conclusion:
"An optimum bias setting exists but it is not specific of the device used but both of the device and the type of circuit in which the device is used".
Is it this way ? Have I misunderstood something ?
In my complete ignorance I though it were specific of the device used, indipendently from the circuit.
I would lke to thank all of you Sirs for these very valuable lessons.
I start to learn, I think.
Regards,
beppe |
|
|
| AndrewT |
Hi,
I believe that the general rule:
that FET output stage benefit more from higher bias in ClassAB than BJT. |
|
|
| beppe61 |
| quote: | Originally posted by AndrewT
Hi, I believe that the general rule: that FET output stage benefit more from higher bias in ClassAB than BJT. |
Excuse me, when you say FET output stages you mean power MOSFETs ?
I have read the MOSFETs work well only when used in class A mode.
Anyway even when used highly biased they tend to stabilize their temperature, so they are quite suitable for class A designs, differently for instance from bjts.
Thanks and regards,
beppe |
|
|
| MikeB |
Hi Beppe, i think you mixed up something about the mosfets. There are 2 different types of mosfets, the expensive lateral mosfets, these are designed for audio and can be well used in ClassAB and have the self stabilizing bias with temperature, and the vertical mosfets (IRFs for example), these are not so suitable for ClassAB and are not thermally stable, but very cheap.
The "An optimum bias setting exists but it is not specific of the device used" is about the fact that the nearly same optimal bias exists for all bjts.
But, if these bjts are too slow, the crossoverdistortion created by their insufficient speed dominates and needs to be biased higher. I personally think of these devices beeing not adequate for use in ClassAB outputstages.
Mike |
|
|
| beppe61 |
| quote: | Originally posted by MikeB
Hi Beppe, i think you mixed up something about the mosfets. There are 2 different types of mosfets, the expensive lateral mosfets, these are designed for audio and can be well used in ClassAB and have the self stabilizing bias with temperature, and the vertical mosfets (IRFs for example), these are not so suitable for ClassAB and are not thermally stable, but very cheap.
The "An optimum bias setting exists but it is not specific of the device used" is about the fact that the nearly same optimal bias exists for all bjts.
But, if these bjts are too slow, the crossoverdistortion created by their insufficient speed dominates and needs to be biased higher. I personally think of these devices beeing not adequate for use in ClassAB outputstages.
Mike |
Thank you Mr. Mike.
So the fundamental statement really is Nearly same optimal bias exists for all bjts being that one that gives about 25mV across RE.
I think it is a very fundamental point, isn't it ?
Actually I am mostly interested in power amps with output bjts having my amp two 2SC5200/2SA1943 couples per channel.
Hoping that are not "fake" Toshibas of course.
I have read a lot about fake Sanken (maybe I am silly but I love Sanken bjts).
The REs are 0,22 ohm.
I am sure the actual bias setting on my amp is much lower that this optimal, remaining the amp completely cold even when left on continuously.
Thanks again and regards,
beppe |
|
|
| MikeB |
I forgot to mention, this 25mv rule applies only to emitter followers using bjts.
The drawback is, with 2 pairs biased at 110ma you have a total of 220ma. Assuming 48v rails you have idle dissipation of 21watts per channel. The question now is, does the amp sound that much better at the higher bias to justify a total idle dissipation of 42watts ?
Building/designing amps is always a big collection of compromises.
Also, many amps are designed in a way that the amp blows up when the wiper in the bias pot looses contact. (Stupidity, not compromise)
When amp is older, the bias pot gets dirty, guaranteeing that the amp blows up when trying to adjust the bias...
Mike |
|
|
| jacco vermeulen |
| quote: | Originally posted by MikeB
this 25mv rule applies only to emitter followers using bjts. |
Big plus again for Sziklai.
There was a time when a Cermet PT10 was regarded as high quality. :clown: |
|
|
| janneman |
| quote: | Originally posted by MikeB
[snip]Building/designing amps is always a big collection of compromises.[snip]Mike |
:D
Jan Didden |
|
|
| Babowana |
| quote: | Originally posted by janneman
:D |
Meaning "yes" or opposite . . . ? |
|
|
| beppe61 |
Originally posted by MikeB
I forgot to mention, this 25mv rule applies only to emitter followers using bjts.
If I am not wrong most of the output bjts are arranged this way.
Am I wrong?
The drawback is, with 2 pairs biased at 110ma you have a total of 220ma. Assuming 48v rails you have idle dissipation of 21watts per channel.
I think that this amount of dissipation requires very large heatsink. So I feel it is not applicable in my case without changing the box. The heatsinks' surface is just normal.
So a compromise is needed.
The question now is, does the amp sound that much better at the higher bias to justify a total idle dissipation of 42watts ?
Nice question. Not easy to answer I guess.
Also, many amps are designed in a way that the amp blows up when the wiper in the bias pot looses contact (Stupidity, not compromise).
When amp is older, the bias pot gets dirty, guaranteeing that the amp blows up when trying to adjust the bias...
I did not know. Very very risky.
Now I understand why some audio "restorers" change firstly trim pots with new and better multiturn ones (and the PS filter caps as well).
I noticed in an old amp very bad trimmer on the PCB.
I do not play with bias any more.
Thanks and regards,
beppe |
|
|
| MikeB |
Beppe, you're right, the emitter follower arrangement is the most common. But there are also:
- Transconductance output
- Quasi complementary (fore sure not the case in your amp) Meeeeow....
- Sziklai (CFP)
- more exotic ones
About the bias pot, as long as it is capsuled, the risk of loosing contact is minimal. If the pot is the cheap open construction, it gets risky. But only some amps have this flaw. Depending on how the vbe multiplier is built, loosing the contact will completely open the multiplier (giving zero bias) or the multiplier completely closes, supplying the full voltage rail as bias. (bang)
Mike |
|
|
| beppe61 |
| quote: | | Originally posted by MikeB |
Beppe, you're right, the emitter follower arrangement is the most common. But there are also:
- Transconductance output
- Quasi complementary (fore sure not the case in your amp)
- Sziklai (CFP)
- more exotic ones
Thanks for the explanation.
About the bias pot, as long as it is capsuled, the risk of loosing contact is minimal.
If the pot is the cheap open construction, it gets risky.
It seems to me that all these pots so important should be of the incapsulated type (I think like the blue Bourns and similar).
They are a very delicate component in an amp and some care should be used.
The old amp I was referring to has indeed open type pots.
Better not to play with it.
Thank you very much again.
Regards,
beppe |
|
|
| janneman |
| quote: | Originally posted by Babowana
Meaning "yes" or opposite . . . ? |
The statement was: "Amp design is a lot of compromises" (or similar). What do you think, after reading stuff on this forum?
Jan Didden |
|
|
| Babowana |
| quote: | Originally posted by janneman
"Amp design is a lot of compromises" |
I have lived, thinking (noy only amp design) that many designs are lot of compromises. I don't mean compromises in quality itself, but in process to get the target quality.
me, time to go to bed . . .aaaah . . . hmm . . . |
|
|
| sam9 |
| "No compromise" is just advetising smoke. There is always somthing, if only cost, mass or practicality. |
|
|
| janneman |
| quote: | Originally posted by Babowana
I have lived, thinking (noy only amp design) that many designs are lot of compromises. I don't mean compromises in quality itself, but in process to get the target quality.
me, time to go to bed . . .aaaah . . . hmm . . . |
Yeah. Life is just another word for compromising. ;)
Sleep well!
Jan Didden |
|
|
| 4fun |
Hi!
beppe 61,
I have seen some internal pics of Behringer A500 and as I can see there is only two supply leads from transformer to each channel (no center winding ground lead), can you comfirm this?
If that is the case, together with Behringers behaviour of using schematics from QSC implies that we may have "flying rail" topology implemented here like QSC uses on several of their amps.
Maybe your schematic is simular to this:
http://www.qsc.com/support/library/...ries/usa370.pdf |
|
|
| beppe61 |
| quote: | Originally posted by 4fun
Hi!
beppe 61, ....
|
Hello Mr. 4fun.
I have seen some internal pics of Behringer A500 and as I can see there is only two supply leads from transformer to each channel (no center winding ground lead), can you comfirm this?
Firstly I declare my complete ignorance in electronics.
Then I asked Behringer for a service manual getting a resounding no. I will try through a service lab anyway. But I feel it will be a difficult task.
From what I can see YES. There are only 2 wires coming out of the transformer and going to each PCB/channel.
They are connected to the PCBs with thermally glued connector.
I do not know how to remove the glue for measuring the AC voltage in the secondaries.
So YES. There are no center winding ground lead.
And this was the first thing that surprised me. A very uncommon scheme.
If that is the case, together with Behringers behaviour of using schematics from QSC implies that we may have "flying rail" topology implemented here like QSC uses on several of their amps.
It is highly possible. Thank you very much for your advice.
The filter caps are indeed 3300uF like in the schematic.
It should be also the same of the Alesis RA300.
Very similar aesthetically.
Maybe your schematic is simular to this: http://www.qsc.com/support/library/...ries/usa370.pdf
Thank you very much again.
I will try to understand if the schematic is correct, maybe with the help of a technician.
I fell this amp has some unexpressed potential.
For now it is hard on bass and harsh in the mid.
Not very pleasant at all.
But the toroidal is large (about 600VA) and the output devices 2 pairs of 15A Toshibas.
I would like to try to add capacitance in the power supply, just to start. But the space on the PCB is minimal (the 3300uF caps have 25 mm diameter and there are a lot of components surrounding them).
Do you have any advice about any mods that could be performed?
Thank you very much indeed.
Kind regards,
beppe |
|
|
| beppe61 |
A picture of the very similar ALESIS RA 300 .
Regards,
beppe |
|
|
| sreten |
| quote: | Originally posted by janneman
Doug Self makes a strong case that it IS applicable to ANY (bjt) output config, be it EF or CFP. It has to do with the relation between the voltage across Re and Vbe.
There is of course a difference in Ibias for the different output stage configuration when the Vre is the same. The EF stage will have almost an order of magnitude larger Iq than the CFP for optimal Vre bias. Your quoted 10mA is Dougs recommendation for Re=0.22 ohms (actually, 11.5mA at Vbias=1.297V).
Jan Didden |
Huh ? "25mV across RE" as a blind rule you show is wrong.
So "a strong case that it IS applicable to ANY (bjt) output config,
be it EF or CFP ", a strong case for what ? What is this "it" ?
:)/sreten. |
|
|
| jacco vermeulen |
| quote: | | order of magnitude larger Iq than the CFP for optimal Vre bias |
25mV : 2.5mV = 10 : 1 |
|
|
| AKSA |
Jacco,
I see your Oz namesake has come in #2 in the MotoGP!
Related perhaps?
Hugh |
|
|
| jacco vermeulen |
Mr Dean,
The main attraction in my family is coming up with an original way to do yourself in.
One of my uncles succeeded in convincing a Samoerai gentleman to cure his headache during the war years in Indonesia.
My very creative kid sister just did drugs and received an A+ by hopping into a coma.
Up till now i've been a blatent disgrace to my heritage.
After a pityfull electrocution attempt i displayed my lack of talent 4 years ago by copying my sister.
Unfortunately my skills are so deplorable that i've only managed to remain a comatose vegetable for a week and a half.
The few historical exceptions i can think of are my grandfather and my great uncle, the latter seems to have been the only classical music composer Belgium ever had.
Gramps definitely tried by smuggling meat during WWII. Thanks to the miscommunication of our occupying neighbours grandpa came out of the war filthy rich and married a baron's daughter.
I regard these two as evolutionary mistakes.
The motorsport routine is a trifle boring, i can think of at least 2 of my family members who took that easy escape route.
Chris drives a FORD truck, can you believe that ! |
|
|
| janneman |
Jacco,
Your family birthday parties must be a riot!
Great story!
Jan Didden |
|
|
| AKSA |
Jacco,
Interesting, colorful story - but we would expect no less.
Wrestling with ones intellect, addictions and curiousity are what makes us human, and I wouldn't be dead for quids..... only two days ago on a visit to Peter Brock home turf on my ZRX I briefly thought I too was at the MotoGP - you really know you are alive when you challenge the laws of physics, or convention, or medicine, or pretty much any rule......
Cheers,
Hugh |
|
|
| jacco vermeulen |
Beppe,
on a side note:
Besides the difference of parallel component number, heatsink capability and bias between class A and class AB output stages there's also a variance in driver stage layout.
Many seem to focus their eyes on the output stage only.
The output stage bias determines the zero crossover point for the output devices, but there's also a crossover point for the driver stage.
At higher output levels not only will the output stage have changed to class B operation, at a lower impedance level the driver stage will also reach the crossover point.
For an example: the output stage of a fully biased KSA is good for full class A operation into 8 Ohms, the driver stage does full class A in 4 Ohms. At higher output levels the output stage reaches class B mode, while the driver stage is still in the class A area.
Mr D.,
A few days ago my g/f and I talked about Stephen Timoshenko after I dropped the father of mechanical engineering's name here. In her opinion hardships are often catalytic converters for brilliance to excell. http://histsoc.stanford.edu/pdfmem/TimoshenkoS.pdf
On the other hand, Q. regretted his utter foolishness on his dying day in Don Q. revisited. |
|
|
| beppe61 |
| quote: | Originally posted by jacco vermeulen
Beppe,
on a side note:
Besides the difference of parallel component number, heatsink capability and bias between class A and class AB output stages there's also a variance in driver stage layout.
Many seem to focus their eyes on the output stage only.
The output stage bias determines the zero crossover point for the output devices, but there's also a crossover point for the driver stage.
At higher output levels not only will the output stage have changed to class B operation, at a lower impedance level the driver stage will also reach the crossover point.
For an example: the output stage of a fully biased KSA is good for full class A operation into 8 Ohms, the driver stage does full class A in 4 Ohms. At higher output levels the output stage reaches class B mode, while the driver stage is still in the class A area.
... |
I think I understand your point: both the driver and output stages must be biased for minimum distortion.
So it seems very sensible to separate as far as possible the two stages.
I remember an amp where the ouput stage had separate supply and acted as unity gain buffer with the whole voltage gain given by the driver stage powered by an independent power supply.
In that way could it be easier to bias the two stages individually?
It seems to me a smart solution: isn't it ?
Thanks and regards,
beppe |
|
|
| jacco vermeulen |
Not exactly, my last post was more than vague.
The switch-off point of an EF driver is at an output voltage that is only a little bit higher than the output voltage at which the output devices go to class B.
Remember that the Vbe multiplier keeps the voltage difference between the bases of the NPN and PNP drivers constant.
If the voltage on the output node raises by 1 Vbe drop the driver stage will also go to class B. |
|
|
| AndrewT |
Hi Jacco,| quote: | | The output stage bias determines the zero crossover point for the output devices, but there's also a crossover point for the driver stage. |
I wish you had read my enquiry on the Leach thread. No one confirmed my suspicion that Leach was wrong, although a few thought he is right. Would you care to comment?
Was Timoshenko a civil engineer? |
|
|
| MikeB |
Something very interesting: http://www.diyaudio.com/forums/show...145#post1009145
Chris (anatech) was adjusting bias in his symasym with a distortion analyzer connected, at 22ma crossoverdistortion was lowest... :confused:
I wish i had such a tool. :bawling:
Mike |
|
|
|
