Consider, if you will the case where we take a PSU on a class B amp (to make it clear) and place an external load on each rail (resistor) that draws some amperage (1.3 amps?).
Now is that the same condition as is that same class B amp's rail voltages when driven?
Assume the same PSU.
It's load vs. no load conditions. Somewhat different in practice.
And, what you said is essentially what I said... but in practical terms for most PSUs modulation of the "current" causes a dip in the available voltage.
_-_-bear
Consider what happens to ClassA amplifier's supply rails when biased to 1.3A and the rail current reduces to 0.1A and then what happens to the rail voltage when the rail current increases to 2.5A.
The supply rail voltages are modulated, just like a ClassAB amplifier that is modulating it's supply rails due to currents varying from 0.1A (bias current) to 2.5A (maximum current to match the ClassA amplifier).
It is in practice somewhat different.
In one case the load is being reduced, and the rail voltage climbs, (depending upon the time constant of the PS to some extent) in the other the load is being increased and the rail voltage drops and also the hum and ripple increases.
In the AB amp and even more in the Class B amp the PS is asked to suddenly supply current from a nil current draw, rather than to vary about a significant quiescent current draw.
I see it as quite a different case, ymmv.
_-_-bear
I disagree.
The output impedance of the PSU interacts with the output current demanded by the amplifier.
Current times impedance equals voltage.
A ClassAB amplifier can change it's supply rail current from 0.1A to 2.5A and back down to 0.1A during a half cycle of the output waveform.
A ClassA amplifier can change it's supply rail current from 1.3A to 0.1A and back up to 1.3A during a half cycle of the output waveform. In the next half cycle the rail current changes from 1.3A to 2.5A and back down to 1.3A.
In both cases the PSU sees a current demand changing from 0.1A to 2.5A and if the PSU impedance is the same then the voltage modulation will be the same.
You seem to be saying that the ClassAB amplifier modulates the PSU voltage and that the ClassA amplifier does not. If that is your assertion then it is wrong.
The output impedance of the PSU interacts with the output current demanded by the amplifier.
Current times impedance equals voltage.
A ClassAB amplifier can change it's supply rail current from 0.1A to 2.5A and back down to 0.1A during a half cycle of the output waveform.
A ClassA amplifier can change it's supply rail current from 1.3A to 0.1A and back up to 1.3A during a half cycle of the output waveform. In the next half cycle the rail current changes from 1.3A to 2.5A and back down to 1.3A.
In both cases the PSU sees a current demand changing from 0.1A to 2.5A and if the PSU impedance is the same then the voltage modulation will be the same.
You seem to be saying that the ClassAB amplifier modulates the PSU voltage and that the ClassA amplifier does not. If that is your assertion then it is wrong.
I am asserting that "modulating" a PS that is at a nil quiescent current vs one that is somewhat heavily loaded quiescent is a different case in the real world.
That they modulate the same absolute value of current is not the point I am making.
But you are free to think of it differently than I do.
_-_-bear
Cascoded version
Hi
I have decided to make another gb run of the f5 pcbs, but instead of making the same again I have also decided to make the boards longer.
But all that extra space made me want to include some cool options for people to try out, which leads me to the question.
Has anyone tried cascoding with jfets as in the attached schematic? I have searched the thread, but only found the cascoding done with bjts.
Hi
I have decided to make another gb run of the f5 pcbs, but instead of making the same again I have also decided to make the boards longer.
But all that extra space made me want to include some cool options for people to try out, which leads me to the question.
Has anyone tried cascoding with jfets as in the attached schematic? I have searched the thread, but only found the cascoding done with bjts.
It's not a good idea to cascode k170 with another k170 - it won't leave enough Vds for lower k170 to work properly. Read the articles on E.Borbely's site, it's all well explained - he regularly uses k246 (high Vp JFET) as cascode. Nelson also gave valuable hints about using a cascode, they are in the thread 😉
Also, as you suppose, you can leave out components drawn in squares.
Thanks juma - great point, thanks! Oh no - my least favourite hobby of all - searching for complementary jfets... 🙁
Yeah, the components in the squares are to be documented very thoroughly because I want people to be able to use either set of power fets or both.
Yeah, the components in the squares are to be documented very thoroughly because I want people to be able to use either set of power fets or both.
Not at all, I often get asked by people if i can supply an amp with a clip LED or bargraph. People want to know if the amp is being overdriven. Overdriving speakers can destroy them.
A bargraph is simple enough, just one chip and a few components.
A clip LEd is just a transistor a and few resistors, either is much cheaper than a new set of speakers.
Coool😉
Thats what I suggested early on in the cascode debate, cascode with matched Jfets, but was told it was a complete waste over using say BC###
As pr Juma, it appears like ordinary paralel Jfets may be your option
Others seem to use cascode BC### with Jfets too, guys who seem to know their stuff well
http://www.amb.org/audio/beta22/
Please, consider to make it easy to bypass(jumper) limiters etc, to do the simple version
But you seem to have thought about that option in schematic
Im not sure sims tell you much really, we have been there before with this, but you know more about that
Thanh is working hard on this issue/subject, and his results might be interesting
We have a working BC### cascode curcuit (Juma), but I dont know whether we have the optimal compensation yet
Last edited:
Hello Cviller,
I think that is a great idea to include cascoding option. I just got the parts to do exactly that, but based on Choky's and Juma's work. I was just thinking how to do it on the separate board in addition to Peter's board, that is so tight and with that hard to modify. I was thinking additional board above. Since I am ready with the parts and have test set up, I would be happy to run any test set up for you in order to check it out. Just let me know, and I will be happy to do so.
AR2
Thanks AR2, I would love to hear about your findings.
Cascoding with jfets appears to be a matter of finding the correct jfets high Vp and high current (according to the borbeley article, which I just reread). The J174 looks kinda nice, but I'm not sure if it has a N counterpart.
The BC option could probably be the most sensible option, but I just like the simplicity of the jfet option.
Cascoding with jfets appears to be a matter of finding the correct jfets high Vp and high current (according to the borbeley article, which I just reread). The J174 looks kinda nice, but I'm not sure if it has a N counterpart.
The BC option could probably be the most sensible option, but I just like the simplicity of the jfet option.
btw, this cacode debate was related to using multiple output devices
And maybe slightly raised voltage, or raised bias
With a single pair and standard 24V rails, I dont know
Have you thought about making room fore another pair of outputs ?
And maybe slightly raised voltage, or raised bias
With a single pair and standard 24V rails, I dont know
Have you thought about making room fore another pair of outputs ?
Last edited:
Ahh, looks good
Im sure it will be popular
btw, LEDs have been mentioned
Maybe a power on LED would be nice, to some
And maybe another that glows if amp is leaving classA, if possible
Personally, I usually dont care much about lights and such, but still....maybe
Im sure it will be popular
btw, LEDs have been mentioned
Maybe a power on LED would be nice, to some
And maybe another that glows if amp is leaving classA, if possible
Personally, I usually dont care much about lights and such, but still....maybe
Last edited:
I've completed one channel using 2 pairs of Renesis J162/k1058. No cascoding or alteration of gate/source/feedback values. It seems to work fine, although I did not match the parts. I was only able to bias it to 1A before hitting 50C on the heat sink (vs 1.3A and 41C with single pair Fairchild parts). There also is a difference of up to 15C between the mosfets on the same side (a result of non-matched parts?). Is this a problem?
On a separate note, on initial power up I measured 21vdc on the output. A careful inspection revealed a broken solder point between feedback R5 and jFet Q1 (I didn't use R7&R8). Probably damaged during mounting on the heat sink (it's a point to point). Scary. Cviller's new board might be just the ticket to try cascoding.
On a separate note, on initial power up I measured 21vdc on the output. A careful inspection revealed a broken solder point between feedback R5 and jFet Q1 (I didn't use R7&R8). Probably damaged during mounting on the heat sink (it's a point to point). Scary. Cviller's new board might be just the ticket to try cascoding.