That's a very nice looking FFT
I'm using a model found in a large library I downloaded some time ago.
Started building the board actually.
Will probably get it done tomorrow
Since I will use a 12-0-12VAC center tapped EI xformer to power this headamp, the LT1083 vregs are not an option.
The vref CCS shunt regs I've built so får can't handle the current demands of this headamp, so I don't know what to use yet. LM3x7 tracking pre-regulators?
Implemented a bunch of changes suggested in the JLH thread on that headamp today. Letting it warm up now to see what kind of offset I get with the trimpots set to mid point.
It's been about 10-15min so far and I get around 428mVdc (slowly declining).
That is with no caps on outputs and DC-servo not connected.
If I trim down DC offset as close to zero as I can when the headamp is warmed up, can a servo handle the initial high DC offset then?
Not using any load at the moment either as I forgot to push the release button on the jack, and the tip of the 3.5 to 6.3mm adaptor came off lol.
It came off in such a way that I could glue it back on though.
There are some serious gremlins in my headamp builds lol.
Either that or I have more bad luck/is more dumb than most
Yes, a servo can easily handle that offset as long as you make the resistor from the opamp output low enough in value.
Have you got the 'book' on opamp distortion by Samuel Groner ? If you go to the 'ICopamps' link at the left and then look for the opamp_distortion file. And its a big download at around 35Mb from memory. Samuel shows real FFT's for all the common devices in use.
SG-Acoustics · Samuel Groner · Analogue Audio
Have you got the 'book' on opamp distortion by Samuel Groner ? If you go to the 'ICopamps' link at the left and then look for the opamp_distortion file. And its a big download at around 35Mb from memory. Samuel shows real FFT's for all the common devices in use.
SG-Acoustics · Samuel Groner · Analogue Audio
I currently have 470K resistors on the servo output.
Thanks for the link
Downloaded it just now.
About the AD797 board, do you think 200pF across input resistors and 200pF in series with 100R across feedback resistor(?) the one from inv.input to output is about right to keep the AD797 from oscillating?
I used 10K input, 20K between inv.input and output.
Does 100R seem like an ok value between AD797 output and diamond buffer?
Lowered supplies to +/-10Vdc as mentioned earlier and the diamond buffer output resistors to 1R55.
The BC's will run a bit cooler and the BD's slightly hotter, the heatsinks should handle that fine though.
100 ohm sounds good for isolating the opamp output to the buffer.
The AD797 is one device I haven't used so I wouldn't like to commit on what might be optimum for it.
Have you studied the data sheet and looked at the examples ? The device works best with relatively low impedance feedback networks. 20k sounds very high.
The AD797 also has a reputation for needing extreme care in its implementation and layout as well.
The AD797 is one device I haven't used so I wouldn't like to commit on what might be optimum for it.
Have you studied the data sheet and looked at the examples ? The device works best with relatively low impedance feedback networks. 20k sounds very high.
The AD797 also has a reputation for needing extreme care in its implementation and layout as well.
Yes I read the datasheet.
Åt first I looked åt 1K input & 1K feedback, possibly with 33pf in series with 100R across feedback resistor.
I don't think that simulated as well though.
IIRC, that would give a gain of 2.
Then I read something I don't know if I understood, source resistance should not be >200kohm/G in unbalanced mode.
Does this mean that it doesn't like high resistance(impenance) source?
Maybe better to go with 33R series resistor and 300R feedback resistor for a gain of 10 and use 5pF caps across feedback and series resistors?
I guess I'll do some more simulations before continuing the build of that board.
Btw, with the changes made to the JLH, it was easier to trim down DC offset.
It seems the DC offset drift was alot less as well.
Might be ready to put in the DC-servo tomorrow, hopefully the 470K output resistor will be ok.
Using 1 Meg as input resistor.
With something like the AD797 you must do most of the design proofing as an actual build and verify the results with real measurement. Relying on simulation for this just isn't going to be a guarantee its OK.
Have a look at this,
http://www.diyaudio.com/forums/anal...rent-feedback-opamps-simulation-practice.html
Thanks again
I'll try the "by the book" datasheet circuit for a gain of 10
R1: 33R R2: 300R bypassed with 5pF in series with 100R
And I'll use the supply decoupling from the datasheet as well.
Hopefully this will work, if not..do it again, do it right as my old commanding officer used to say.
That looks nice
Yes, that was my thinking as well.
I use the solder station so much, so I figured a simple one from a good brand makes the most sense.
Well I've decided to try AD797's as well.
I will re-work the diamond buffer board to fit 100R trimpots on two of the four CCS's and swap the output resistors for parallel carbon film 3R3.
I'm nearly done, enough for testing with DMM anyway, with the AD797 board.
Lytics are Panasonic FC 4.7uF, they're parallelled with 100nF 0805's.
I've added 22uH in series with the powersupply as that atleast made a rather big difference in simulations.
It looks messy, but it'll look better once the flux etc is cleaned off
I've got a few more wires to solder in and then it is ready for some extensive testing with the DMM making sure that there are no shorts etc before I try it with the diamond buffer.
I hope I've gotten the PS decoupling close enough, and good enough, to prevent the AD797's from oscillating.
I will re-work the diamond buffer board to fit 100R trimpots on two of the four CCS's and swap the output resistors for parallel carbon film 3R3.
I'm nearly done, enough for testing with DMM anyway, with the AD797 board.
Lytics are Panasonic FC 4.7uF, they're parallelled with 100nF 0805's.
I've added 22uH in series with the powersupply as that atleast made a rather big difference in simulations.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
It looks messy, but it'll look better once the flux etc is cleaned off
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
I've got a few more wires to solder in and then it is ready for some extensive testing with the DMM making sure that there are no shorts etc before I try it with the diamond buffer.
I hope I've gotten the PS decoupling close enough, and good enough, to prevent the AD797's from oscillating.
Well, I don't have one unfortunately.
All parts are datasheet values, except for the added inductor...one on each rail and op amp.
All parts fitted on a 4*6cm board.
I have 15pf in series with 100R across the FR(300R).
200R to GND, 150R input resistor and 82R between op amps and diamond buffer.
When dealing with wide bandwidth parts its the layout as well as component values that plays a big part. For example wiring and PCB runs start to become inductive at high frequency and can actually form 'oscillators' in combination with the active devices on the PCB.
Its not easy to explain... a short piece of wire is anything but a short circuit at high frequency.
Its not easy to explain... a short piece of wire is anything but a short circuit at high frequency.
I get it =)
I'll try it, should it not work I'll try some other opamps (changing required parts accordingly).
Hmm, maybe swap signal wiring(perf board "traces") for thicker Cu wire might be a good idea? Used the roll closest (0.2mm) as I use that to easily create "traces" on perf board but I have 1mm thick wire as well.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.