nelsonvandal said:
optimum Cn from my simulations looks to be very "tweaky", you are essentially canceling the internal Ccomp and any device parasitic C at that internal node – some of which may be V dependant, hopefully at least the bulk Ccomp is a MOS cap
and the cancellation occurs through the current mirror which likely has some tolerance on current transfer ratio and a high frequency roll off
90% cancellation leads to a 10x greater loop gain around the output stage-Cn local feedback loop and corresponding distortion and output impedance reduction
to get to the ~200x max boost I found in my subsequent sims of the simplified datasheet circuit in the real world seems problematic - you could try to measure the output Z as you trim the Cn cap but you'd need sub mOhm resolution at ~100 KHz - a daunting measurement with nH lead inductance adding ~mOhm/inch of lead wire impedance - even varying position of the gnd ref probe point on a full 2 oz ground plane will make a difference at these levels
Nice use of a Howland current pump JCX.
Scott, I seem to recall you mentioning that modern NPN and PNP IC process transistors really performed well and class A biasing was not neccessary to get good performance. I guess thnkgs in this area hav e changed a lot sinnce the late 70's and early 80's.
Scott, I seem to recall you mentioning that modern NPN and PNP IC process transistors really performed well and class A biasing was not neccessary to get good performance. I guess thnkgs in this area hav e changed a lot sinnce the late 70's and early 80's.
Bonsai said:
Well they are now much more complimentary, but as I keep telling JC the world wants low power so most general purpose amps are fairly light on the bias. We have some performance at all costs differential A/D drivers that are different but they are all +-5 or less volts.
john curl said:
Well, you can't
at least not without getting some other severe issues/problems around you neck
An IC, however good it is, is a 'rather' closed system of optimized semiconductors and passives
and AD797 is one of the best no doubt
I would say, it is enough difficult to accomplish a removal of the nasty 7th
in a completely open system = discrete devices
... but this you already know very well Mr. Curl, by experience
I take your enquiry as being a highly retorical one.
lineup 🙂 discrete hifi regards from the audio basement
And on the thread topic, a good amp like the AD797 might only benefit from class A bias under extreme and suboptimal conditions for it's supply and bypassing by making the load induced supply ripple more signal related that it is in class B (half wave rectified portions).
- Klaus
- Klaus
It is best Scott, to ask the actual 'designer' of the AD797 before proceeding on my own. I WILL find out, one way or another, or change to a different part.
john curl said:
What circuit, what frequency, what load, what level? Not a very well "engineered" question. If -130dB is only good for rock and roll so be it, I'm not of that school.
Now that I have your attention: about 850 ohm load at high frequencies, perhaps 1500 ohm below 1KHz. It is next to useless to specify things, until you are willing to help.
Scot, one optiopn to consider, if you have a spare pin, is to be able to program the output Iq via an external resistor to ground, say. Something for the op-amp road map.
Clearly an upper Iq limit needs to set internally.
Clearly an upper Iq limit needs to set internally.
Bonsai said:
I think the days of sneaking those "features" past marketing are over. The audio stuff has all moved to the left coast anyway.
BTW I have family in Ome and my daughter is thinking of possibly relocating sometime. I just got back from Osaka and Tokyo it seems like a bargain compared to my last trip to France $$
$$
All that I ask is what current should I set a current source, connected to -15V and the output, if there is no other approach. I have already designed the circuit, have boards, ready for release, except for this, and similar factors. I have always been plagued with a residual 7th harmonic, even when using AD797's in my ST. Maybe there is a solution. Increased 2'nd or 3'rd isn't very important in my case.
john curl said:
My measurements were at the -130dB level at 600 Ohms, is that really not good enough?
Hi,
i know that the AD797 is one of the best audio opamps
and there are no way to get better results as the normal one
measured. if the circuit is good calculated you don´t need to modify
anything or to try to bias in class a (not with the AD797).
@ scott
which type of analyzer do you use ? or by soundcard ?
i know that the AD797 is one of the best audio opamps
and there are no way to get better results as the normal one
measured. if the circuit is good calculated you don´t need to modify
anything or to try to bias in class a (not with the AD797).
scott wurcer said:
@ scott
which type of analyzer do you use ? or by soundcard ?
Is this with 30dB forward gain? That was the 7th harmonic only, or 'everything'? Perhaps 1-2Vrms output. 5KHz? 10KHz?
Scott, based on fig 14 of your data sheet, I would predict -100dB THD at 1V-2V out. 7th will, of course be less, but probably still measurable. I don't like measurable 7th. It makes me uneasy. Please understand, people expect me to do my best, not 'good enough'.
ECBLN said:
My original measurements were made with a ST and external L/C filters. We are now going back to this approach since the instrument makers have not kept up with the capabilities of our puny IC's. We now use a Stanford Research diff out source.
- Status
- Not open for further replies.