I was being serious. There are a few basic 'circuit blocks' and these are combined in new ways or refined over time and eventually end up SOTA. Certainly some components have improved over time either due to design, or process improvements in th e semi industry, and this helps move things forward, others to my mind are a step back (anybody for class D?).
I've been trying the folded-cascode in various guises and get the feeling it is quite a fiddly topology.
I getting more convinced by the day that for line level signals, CFA is the way to go.
Looks like th e folded cascode also has PSU noise rejection problems, but not as bad as CFA.
I guess its a case of Pick your poison.
I getting more convinced by the day that for line level signals, CFA is the way to go.
Looks like th e folded cascode also has PSU noise rejection problems, but not as bad as CFA.
I guess its a case of Pick your poison.
Glen,
I`m glad to see your unexpectedly arisen interest in this subject. Can the summary of investigation be summarized yet?
I`m glad to see your unexpectedly arisen interest in this subject. Can the summary of investigation be summarized yet?
Well, the folded cascode "VAS" can be used in a CFA - It will work just the same except that the slew rate will be limited as, unlike a current mirror stage, it can't deliver several times the standing current to the compensation capacitance.
But why does the folded cascode have PSR issues?
But why does the folded cascode have PSR issues?
The PSU rejection on the CFA is not good (I injected some signals into the PSU rails to look at this - simulation of course). The standard symmetrical VCA with LTP and mirror loads is very good. I did some simulations on the folded cascode some time ago, and I got a better result than the CFA, but a long way off the VCA. I am trying to find the original sims I did.
Yep, pretty extensively simmed and built the Haksford cascode, Baxandall
pair and a self improvised X coupled darlington arrangent.
They all do pretty much the same thing - re process the base current and
non linear bc capacitance. They are all susceptable to oscillation, especially
the Baxandall pair, probably because the 'clean up' transistor works with
very little head room. In the event of oscillation, decoupling the 'clean up'
transistors base to ground at VHF usually works but not always.
I like this topology applied to a power amp but unforunately when you
substitute that ideal infinite current gain OP stage on your sims for a real
one, all those really cool distortion numbers get shot to pieces.
Then the design task becomes more - what OP stage can I put with the
Hawksford VAS to preserve it's merits. This becomes a much more tricky
task.
Good thread Glen.
cheers
Terry
Glen,
post #1 hardly contains a single correct statement, just a large number of disturbing assumptions.
The beta is the ratio of the change in collector current to the change in base current. It is not a functional relationship. Currents don`t control currents. Also, there is no direct relationship between the beta and voltage gain.
The alpha is the ratio of the change in collector current to the change in emitter current. This ratio is 0.98, setting the current gain of the grounded base amplifier. The alpha is of diminishing functional importance.
Likewise, regarding the Hawksford paper (J10), nonlinearities and slope distortions cannot be explained by the alpha factor.
In the grounded base amplifier does not occur any Miller amplification.
Now, what is all that talk about hfe?
post #1 hardly contains a single correct statement, just a large number of disturbing assumptions.
The beta is the ratio of the change in collector current to the change in base current. It is not a functional relationship. Currents don`t control currents. Also, there is no direct relationship between the beta and voltage gain.
The alpha is the ratio of the change in collector current to the change in emitter current. This ratio is 0.98, setting the current gain of the grounded base amplifier. The alpha is of diminishing functional importance.
Likewise, regarding the Hawksford paper (J10), nonlinearities and slope distortions cannot be explained by the alpha factor.
In the grounded base amplifier does not occur any Miller amplification.
Now, what is all that talk about hfe?
Terry, I thinky you can pretty much say this about any VAS.
I am not sure I fully understand your point here though. I am feeding a Hawksord VAS from a current mirror and the VAS feeds a diamond buffer. I am loading the VAS in order to vary loop gain independent of closed loop gain and I get some pretty good figures.
What exactly is 'shot to pieces'? If its 500ppb with 40db of fb at 20KHz to 3.5ppm with 6db of fb at 20KHz then I agree with you its 'shot to pieces'. Otherwise , its pretty sh1t hot in my view - even at 3.5ppm distortion at 20KHz.
Please define 'shot to pieces'
Last edited:
Just to claify my statement ' ' . . . you can pretty much say this about any VAS'
Using a standard VAS typically yields distortion figurs of 1 to 2 orders of magnitude higher than the H-VAS (I'm talking 20KHz THD here).
When then loaded with driver and output stage, this makes he situation worse. Thats why most folks go for a triple which helps somewhat.
Using a standard VAS typically yields distortion figurs of 1 to 2 orders of magnitude higher than the H-VAS (I'm talking 20KHz THD here).
When then loaded with driver and output stage, this makes he situation worse. Thats why most folks go for a triple which helps somewhat.
Well, at least you are working on a quality VAS, and not just something thrown together to make some voltage swing, or else just an IC. Looking at virtually every IC made today for audio or video, I still don't find a nearly ideal device that works really as well as a well designed discrete design.
While a triple works OK, I prefer the combination of complementary mosfet-bipolar output. It is linear, fast, hi Z, and it helps us win a lot of awards.
While a triple works OK, I prefer the combination of complementary mosfet-bipolar output. It is linear, fast, hi Z, and it helps us win a lot of awards.
The combination of driver Mosfets and bipolar output is very good. The high gain of the Mosfets reduces consumption of VAS and distortion is very low...
Bonsai,
I am talking power amplifier here not pre amplifier.
They need to be able to handle difficult loads and exhibit good HF behaviour.
My usual design criteria is 4R / 200W / 20kHz, ie 40V pk. And I'll usually
throw in a 1 ohm load at full and 1/2 swing just to see what the topology
does in a major stress condition. Many speakers go well below 4R.
Given the above load conditions, it becomes much more difficult - but not
impossible - to take advantage of the Hawksford VAS stage low distortion
characteristics.
cheers
T
Terry - ok, I understand your statement.
My experience (simms and distortion measurment on a practical amp) indicate that once you start driving very low impedance loads (1 Ohm - very painful!), there are lots of issues to deal with - even if the design employs a triple.
John,
I have no expereince with mosfets in the VAS, but I imagine they have some very useful properties in the type of circuit we are talking about here. Mind you, watch out for the gate capacitance - it's quite a load to drive at higher frequencies.
My experience (simms and distortion measurment on a practical amp) indicate that once you start driving very low impedance loads (1 Ohm - very painful!), there are lots of issues to deal with - even if the design employs a triple.
John,
I have no expereince with mosfets in the VAS, but I imagine they have some very useful properties in the type of circuit we are talking about here. Mind you, watch out for the gate capacitance - it's quite a load to drive at higher frequencies.
- Status
- Not open for further replies.