Electronics > Projects, Designs, and Technical Stuff

Analog challenge:High power/high speed amplifier for physics experiment

<< < (4/6) > >>

dmills:
That spec look a LOT like the design Win Hill shows in the X chapters (and that is discussed extensively over on DIY audio) :

https://www.diyaudio.com/forums/solid-state/287023-winfields-100w-dc-10mhz-1000v-amplifier-12.html

100V (Not 1kV), 5A, 5MHz... Impressive.

langwadt:

--- Quote from: dmills on February 23, 2020, 01:10:23 pm ---That spec look a LOT like the design Win Hill shows in the X chapters (and that is discussed extensively over on DIY audio) :

https://www.diyaudio.com/forums/solid-state/287023-winfields-100w-dc-10mhz-1000v-amplifier-12.html

100V (Not 1kV), 5A, 5MHz... Impressive.

--- End quote ---

he posted a schematic in sc.electronics.design, https://www.dropbox.com/s/elhs1whl2dfn0i7/AMP-70A-2_sch.pdf

David Hess:

--- Quote from: Mazo on February 21, 2020, 04:57:36 pm ---Powered by 30-100V(+- so 60-200V)
Output current up to 4-5A
--- End quote ---

So it needs roughly audio amplifier impedance and power.


--- Quote ---Now the hard part-it should do all that @0Hz-1MHz output frequency(pure sine).
--- End quote ---

Many audio power amplifier designs have almost the bandwidth required but not the gain.  The current feedback design I experimented with before I knew what I was doing achieved 500 kHz of bandwidth with 2 MHz output transistors.  It used current feedback but with a common-base pair at the input instead of a diamond buffer so had no common mode rejection in the driver and output stage.


--- Quote ---Classical voltage feedback discrete amplifier as in a audio amplifier-out of the question as required GBWP is in the order of 10-100MHz :phew:
--- End quote ---

Nothing precludes using a current feedback design.  Also, a voltage feedback audio amplifier would be decompensated to operate at a minimum gain anyway yielding a higher gain-bandwidth and slew rate.  Or a unity gain stable feedback stage might control a fixed gain amplifier to multiply the gain-bandwidth and slew rate.


--- Quote ---Open loop amplifier(Harold Black doesn't like this  :D ) -Hard to get linearity out of?
--- End quote ---

Oscilloscope CRT amplifiers operate like that.  The vertical CRT amplifier is usually just a transconductance amplifier with series feedback for high bandwidth and tame response but the horizontal CRT amplifier also includes shunt feedback to improve linearity and typically only has a bandwidth or 1 or 2 MHz.


--- Quote ---Current feedback topology of some kind using either discrete only devices or a "boosted" op-amp:
--- End quote ---

I would probably try to scale up the unconventional voltage feedback operational amplifier boosted by current feedback into its output which is what your example shows.  But I am not sure it would be up to handling the crossover distortion; I think something would need to be done to control the output stage biasing.

Marco:
I wonder if a fast optically isolated floating driver to drive the output push-pull transistors wouldn't be easier to get high bandwidth with than having to ping-pong from the rails for voltage amplification.

Mazo:
Been busy lately but thought about the design alot,
Reading all of them I thought I should go for a different approach,so I came up with this:


Basic folded cascode which gives me a push pull gain node rather than a CCS loaded one and one gain stage should make for a easy compensation of the amplifier.
The thing which buggers me with the  design:the KSC/KSA trannies are operated very near their limits both powerwise and barely fitting in the SOA->cooling should be aboslutely superb.The high quiescent currents are all in order to obtain a 500-1000V/us Slew rate (I have estimated 30-50pF input capacitance of the output stage).
 The input bias current is also rather large which makes for a hundreds of mV output DC offset(I guess I will have to swallow that one?)
Emitter follower output  makes for easier biasing(Vbe multiplier should sit in the place of Vbias->is ommited for clarity)
The R38-R39-C3 trio should be adjusted IRL and should bypass the output stage(presumed slower than the folded cascode part) at high frequencies to prevent excessive phase shift.
I have calculated and adjusted alot and this design seems to stand a chance,any obvious shortcomings that you see?
Thank you for your contributions.
P.S Alot of needed stuff is omitted for clarity so the schematic doesn't get cluttered,should I include in the real circuit base resistors for all the emitter followers(especially I am afraid the pretty fast 2SC3519A/2SA1386A+the large input capacitance of a 2SC5200/2SA1943 is a recipe for VHF disaster  ;D )

Navigation

[0] Message Index

[#] Next page

[*] Previous page

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod