I am old fashioned too.In some way it reminds me of that old-fashioned accountant who got a computer from boss. As he didn't trust the machine, everyday he checked the results by hand.
I had to learn, at school, how to use a slide rule and then how to use it reliably in my professional capacity as a designer. This of course was before computers became commonplace.
Slide rule calculation demands having an idea of what order of size the answer should be. Yes, we add and subtract the decades to adjust the final answer, but we virtually know the answer before we pick up the tool.
That mindset is applied to all computer modeling.
Hi Edmond,
I agree completely, and my remarks were limited to conventional driver arrangements. Indeed, I proposed the use of ThermalTrak diodes in the driver emitter circuits as well, quite a while back in the ThermalTrak thread. Such arrangements, greatly reducing the DC resistance between the P and N driver emitters, may often need some adjustment from unit to unit to keep the driver standing current at the desired nominal value, however.
The use of push-pull drivers for each of the top and bottom halves of the output stage is also an effective approach, but may carry with it additional cost in complexity and may not end up improving things much unless you allow those push-pull drivers to enter class B (which of course is still better than allowing the output devices to not be turned off adequately quickly)
Cheers,
Bob
Hello Bob & Edmond,
I am very interested in the functionality of these 'speed-up' capacitors. Bob mentioning them in his book sparked my concern/interest.
I have personally used them in 'real life' designs, because in most cases they reduced 'static' THD20 on spice. When it came to calculating the value of them, I thought to myself that the time constant should be very long, so that audio signals are unable / unlikely to change the value of voltage across the capacitor. Therefore, I usually use a 100uF cap.
Are there any conditions that either of you can think of, where they could cause gross distortion, possibly due to severe underbias, e.g. suddenly changing from large signals to small signals (or vice-versa)? If so, I'll try to set up a simulation, where the signal suddenly changes level or frequency.
I look forward (to your undoubtably informative) replies.
Thanks,
Ian
BTW Bob, I have enjoyed reading the kindle edition of your book. You are to be congratulated for adding something significant to the art/science of power amplifier design.
I am very interested in the functionality of these 'speed-up' capacitors. Bob mentioning them in his book sparked my concern/interest.
I have personally used them in 'real life' designs, because in most cases they reduced 'static' THD20 on spice. When it came to calculating the value of them, I thought to myself that the time constant should be very long, so that audio signals are unable / unlikely to change the value of voltage across the capacitor. Therefore, I usually use a 100uF cap.
Are there any conditions that either of you can think of, where they could cause gross distortion, possibly due to severe underbias, e.g. suddenly changing from large signals to small signals (or vice-versa)? If so, I'll try to set up a simulation, where the signal suddenly changes level or frequency.
I look forward (to your undoubtably informative) replies.
Thanks,
Ian
BTW Bob, I have enjoyed reading the kindle edition of your book. You are to be congratulated for adding something significant to the art/science of power amplifier design.
Hi Ian,
Thanks for your kind words about my book. I didn't even know that it was now out on Kindle; cool!
I think that it is unlikely that the speedup capacitor could cause GROSS distortion. However, when we are concerned about every little thing in high-end audio, we just need to be cautious of the use of things like speedup capacitors. If the capacitor is really large, the amount of voltage change it will undergo due to a high-current rate of change event will be smaller (and presumably less damaging/disturbing), but whatever effect it has will last much longer.
As you know, I'm always interested in phenomena that may cause otherwise good-measuring amplifiers to sound less good in the real world with real program driving real loudspeakers.
Cheers,
Bob
Hi Bob,
Another thought I've had: We are all seemingly happy bypassing the Vbe multiplier; might that not have the same similar deleterious effect, in that the voltage across the vbe bypass cap sets the quiescent current of the driver stage via the 2 driver Vbe junctions?
I bought the kindle version, because I gave up waiting for amazon.co.uk to get paperback stock. Incidentally, you dont need a kindle to view it - there is a 'kindle viewer for pc' application downloadable for free from amazon's website. The Kindle edition is cheaper too. Of course, the day after I purchased the kindle version, they got paperback stock!
Another thought I've had: We are all seemingly happy bypassing the Vbe multiplier; might that not have the same similar deleterious effect, in that the voltage across the vbe bypass cap sets the quiescent current of the driver stage via the 2 driver Vbe junctions?
I bought the kindle version, because I gave up waiting for amazon.co.uk to get paperback stock. Incidentally, you dont need a kindle to view it - there is a 'kindle viewer for pc' application downloadable for free from amazon's website. The Kindle edition is cheaper too
. Of course, the day after I purchased the kindle version, they got paperback stock!
Last edited:
Hi Bob,
Another thought I've had: We are all seemingly happy bypassing the Vbe multiplier; might that not have the same similar deleterious effect, in that the voltage across the vbe bypass cap sets the quiescent current of the driver stage via the 2 driver Vbe junctions?
I bought the kindle version, because I gave up waiting for amazon.co.uk to get paperback stock. Incidentally, you dont need a kindle to view it - there is a 'kindle viewer for pc' application downloadable for free from amazon's website. The Kindle edition is cheaper too
Hi Ian,
Your observation about the effect of bypassing the Vbe multiplier is very astute. However, the Vbe multiplier does not go through such large current transitions that might drive its BJT current to near cutoff, as does happen with driver transistors when they are called upon to turn off an output transistor quickly.
Cheers,
Bob
Hi Bonsai,
Yes, that's basically what I do. In the Locanthi Triple I like to run the drivers sometimes at 50-60 mA. For an amp with +/- 64V rails under no-load, this amounts to about 4 watts in each driver transistor, necessitating some non-trivial heat sinking unless they are mounted on the main heat sink along with the power transistors.
For some thermal arrangements, I prefer not to mount the drivers on the big heat sink, instead mounting them, the pre-drivers and a Vbe multiplier on a somewhat isothermal bar on the circuit board. This bar then must be able to dissipate 8+ watts without getting overly hot. It may thus need some heat-sinking enhancement. A separate Vbe multiplier is then used for temperature compensation of the output transistors (often in combination with ThermalTrak tracking diodes).
Cheers,
Bob
There can be a problem with some topologies e.g. the Blameless, that the VAS transistor switches off during clipping, allowing the voltage across the Vbe multiplier to collapse.
"There can be a problem with some topologies e.g. the Blameless, that the VAS transistor switches off during clipping, allowing the voltage across the Vbe multiplier to collapse. "
1. who cares - you're in clipping!
2. however, if it is a concern, try a large decoupling cap actoss the Vbe spreader - I usually use 22 or 33uF
1. who cares - you're in clipping!
2. however, if it is a concern, try a large decoupling cap actoss the Vbe spreader - I usually use 22 or 33uF
clipping
I do care! I'm not only concerned what happens during clipping, but also what happens thereafter. Recovery, you know. Masking the effect with a big fat cap is not the way to go. You better keep the VAS always conducting (and take care of a better clamp).
Cheers,
E.
Hi Bonsai,
I do care! I'm not only concerned what happens during clipping, but also what happens thereafter. Recovery, you know. Masking the effect with a big fat cap is not the way to go. You better keep the VAS always conducting (and take care of a better clamp).
Cheers,
E.
Hi Godfrey,
I hadn't thought of this, but you are certainly right about the mechanism. Let's say the VAS is biased at 10 mA and the Vbe multiplier is bypassed with 10 uF. Let's say that a 5 kHz transient sinusoid is clipped for 1/4 of its cycle, or 50us, collapsing the VAS standing current for 50us. By my calculation, the bias spreading voltage will fall by 50mV, nearly the same amount of quiescent class-AB standing voltage from emitter-to-emitter (i.e., ideally 52mV), causing very substantial transient under-bias of the output stage.
Mitigating this effect, however, is the fact that as soon as the amplifier comes out of clipping the resumed 10mA of VAS standing current will fairly quickly restore the correct Vbe multiplier voltage (I think).
Once again, recognize that if we make the Vbe multiplier bypass capacitor larger, we lessen the magnitude of the effect, but lengthen its duration.
Bear in mind that if 1/4 cycle of a long-period bass transient clipped the amplifier, the spreading voltage might go all the way to zero, and any mid-band program content would be distorted for some time after.
These sorts of amplifier disturbance scenarios are what we want to avoid.
Cheers,
Bob
Hi Bonsai,
Thanks, that's interesting, although it might depend on some details. For example, what part of the amplifier runs into clipping first - the VAS or the output stage.
Another thought here: It may be the case that the use of Baker clamps will prevent the collapse of the bias spreader in most designs. Just thinking out loud here, and I could be wrong.
Cheers,
Bob
I'm not being totally facetious when I suggest using some kind of clipping/overload detector/indicator, and advise one and all to just NOT CLIP THE AMP! I used to just connect a scope across the output terminals, which was quite instructive and caused no sonic ill effects. Much of the discussion I see around amplifier design is about having amps with sufficient clean power output to *never* have clipping. It's obnoxious, it's dangerous, and most of all, it's not good for your ears.
Now if you are folks who are not listening at home and are doing monitoring in studios, etc, and the head-bangers who want to hear what it sounds like when it's really loud are paying the bills, then some form of unobtrusive hearing protection is a really good idea. Having been there, I recommend it highly.
However, if you;re listening at home, perhaps your solution will be to use speakers with decently higher voltage sensitivity/efficiency, so that your megawatt monsters just don't ever have to work that hard.
Really. End of rant.
Now if you are folks who are not listening at home and are doing monitoring in studios, etc, and the head-bangers who want to hear what it sounds like when it's really loud are paying the bills, then some form of unobtrusive hearing protection is a really good idea. Having been there, I recommend it highly.
However, if you;re listening at home, perhaps your solution will be to use speakers with decently higher voltage sensitivity/efficiency, so that your megawatt monsters just don't ever have to work that hard.
Really. End of rant.