General Preamp Design and help selecting opamp(long)

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am set on building a preamp based on a newer high speed opamp because they are one of the newer technologies that might be worth something. I was originally going to use the TI THS6012 based on reading ESP's article about high speed opamps. I have learned of two more units that I think might work with what I am doing. They are the TI THS4012/4022, and the Intersil FHA1100. The former is very similar to the 6012 but is a voltage feedback amp. The FHA1100 is interesting because Intersil claims it to be the fastest op amp available from a semiconductor company (in the datasheet dated 1994, it is significantly faster than the TI products though) It is also the only one available in a regular Plastic DIP package, and also in mil spec'd CERDIP) (is there any advantage to the mil spec stuff other than vanity?, I don't think i'll be boiling water with my preamp.)

Intersil FHA1100
BW: 850 mhz (Av=1)
slew rate 2300V/us
.01dB flatness: 30MHz
.14dB flatness: 100Mhz
Input Impedance: 25k ohm
Max Supply rail voltage: +/- 6V
Output Current: 60mA @RL=50 ohm
Min Stable gain: 1
2nd harmonic distortion @ 30mhz 2Vp-p output: -56 dBc
3rd harmonic distortion @ 30mhz 2Vp-p output: -80 dBc
PSRR: 50dB

Specs on THS4012
3dB BW: 350MHz (Av=10)
Slew rate: 475V/uS
.1dB flatness: 17MHz
Input Impedance: 1M ohm
Max Supply rail Voltage: +/- 16V
Output Current: 100mA @RL=20 ohms
Minimum stable gain: 10
2nd harmonic distortion @ 10mhz 2Vp-p output: -58 dBc
3rd harmonic distortion @ 10mhz 2Vp-p output: -82 dBc
PSRR: 95dB

Specs on THS6012
BW 315MHz
Slew Rate: 1300V/uS
.1dB flatness: 40MHz
Input Resistance: 300k ohm
Max Supply Rail: +/- 16V
Output Current: 500mA @ RL=25 ohms
Minimum stable gain: Not listed, but is stable > 1 and probably at 1 as well.
2nd harmonic distortion @ 10 mhz 2Vp-p output: -59 dBc
3rd harmonic distortion @ 10 mhz 2Vp-p output: -59 dBc
PSRR: -74dB

That is it...

Comparing noise figures, the intersil device doesn't provide any THD specs below 30mhz, so i can only guess they will be really low in the audio band. TI doesn't give any THD figures above 10Mhz. Notice that the intersil device shows comparable distortion figures at 30MHz to the TI devices at 10MHz. The distortion was rising rapidly on the TI devices at 10MHz too. This isn't too much of a problem since in the audio band, the vast majority of test equipment isn't going to be able to get much of a reading on any of the parts. This really isn't the point of my project though. If there are improvements to be made however inaudible I would like to include them unless something more important is needed.

what is a dBc? I have only ever used dBm's, and dBu's. (somewhat related question is what is a is dBA, I see this listed for the sound level of fans...what reference pressure does the A refer to? Something higher than 2x10^-5 so companies can cheat?)

The Intersil device looks like a better part except for a few glaring deficiencies. First being the input impedance. A buffer is needed for sure if I want any sort of volume control(I do). What can I do about this? ESP uses the OPA134 to act as an input buffer in his THS6012 preamp. Will this not destroy the superior noise properties of the CFB amp? Would something like the AD BUF04 be better suited for this? I have never seen it drive another op amp... I am fairly new to this and the effects of the impedance from a previous stage on a latter stage are new to me. Most of my analog design has been in RF where you learn 100 different ways to match the impedances. What are the basics here in audio? How do I choose a volume control pot i.e what will a 10k pot give me that a 100 or 500k pot wont, and vice versa? If I was using the 6012, would I be able to get away without a buffer if the volume pot was only 10 or 20k ?

The second problem is the +/-5V supplies required on the intersil device. This isn't much of a problem I don't think since I was planning on a gain of 2-3 anyways. a gain of 2 wouldn't allow the circuit to clip until about 75% volume with a 2Vrms input. My amplifier has an input sensitivity of about 1.9V anyways. Would be nice to have the headroom though.

My 3rd concern is the power supply rejection ratio. I only bring this up because the intersil part's is so much lower than the other two. Since it only needs +/-5 V, I can probably use really nice, big 16V filter caps, and LT 317/337 with good decoupling. I am thinking 20,000uF-40,000uF or so from nichicon or black gate, or 240,000uF from mallory for the same price. What would I use to decouple something like this?

I am pretty sure ceramic bypass caps are a must, and almost every design i've seen suggests 100nF to bypass with. Would i benefit at all by adding a metal film cap to the mix? say, 100uF electrolytic, 1uF film, .01uF ceramic? This would clutter the board greatly where good close contact is essential. There are certainly different grades of electrolytics and film capacitors, is there as much of a selection with ceramics or are they all pretty much the same? I know the bypass capacitors need to be close and their leads need to be as short as possible. My immediate thought is to drill the PCB holes a bit bigger(If I end up with a DIP), coat all leads with flux and stick them through the same hole, then solder from the bottom trying to get as much solder as possible in the hole so the leads can be clipped off as close to the board as possible. Will this work? I am going to use the eval board for my first attempt, and if it turns out well, I have enough of a need to warrant having some PCB's made with the rest of the circuit incorporated. I am thinking a DIP package might ring like a bell if I try to use the IC to its fullest. Comments?

Instead of a traditional supply, I could use 2 or 4 8.4V NiMh or NiCd batteries and incorporate an auto charger. There is even room to regulate, would that be a total waste with batteries, or does it have some merit? I don't know much about battery supplies.

TI's VFB amp looks nice except for the minimum gain of 10. 20dB of gain in a preamp is not something I think I want. It looks like it would work good as a headphone amp though.

If it is reasonable to use the THS6012 without an input buffer, then I think that is what I would like to do in order to avoid more complexity, cost, and still have a very high performance design. If I would need to use a buffer, then I think I would rather use it with the Intersil part, but again it becomes a tossup between the ths4022 with 20dB of gain, or the intersil + buffer setup.

Ideas? Comments? I told you it was going to be long!

Thanks a lot

jt
 
First thing to do :

Check noise figures. The THS4012 have noisedensity wich is a lot bigger than opa2134 (Only for comparison!!). It flattens out around 1KHz where a device like the opa2134 does this around 100Hz.
But THS4022 looks like a good choice... But you still have to look at the noisefigures:

If you got a low nV/Squareroot(Hz) you got a high pA/Squareroot(Hz). They are normaly inverse.

Try look at the datasheet for an AD797. There is some idears about noise figures.

Texas has some documents about opamp design. They are called "OPAMP's for everyone".

Maybe you should start there.

Sonny
 
jteef,
Regarding the PCB question--I've seen a number of units over the years that do pretty much what you're describing, some commercial, some homebrew, some cheap Japanese years ago, some fairly high-end comparatively recently. It works pretty well as long as you don't overload the hole with too many leads. Some people go so far as to call it point-to-point wiring, with the PCB lands standing in for solder lugs.
Ceramic caps should be viewed with suspicion. Film caps, although bulky, are your best choice. Electrolytics are sometimes necessary due to the amaount of capacitance needed.

Grey
 
Geoff is right about ceramics as bypas of an opamp. If you don't want your highspeed opamp to oscillate (You are a bad OPAMP ... you should be shamed... ), you have to go for ceramics in order of 100nF bypassed by a NPO type of ~ 1nF - 10nF.
Its is all a mather of lead inductance... Place some SMD 100nF ceramic caps no more than 3 - 5 mm from the supplypins and use as wide tracks as possible!
If it is possible make a groundplane.

Sonny
 
Valkyrie

There is an old design from when AudioXpress was the old Audio Amateur Called the Valkyrie. It used AD744jn and AD 811an devices in the circuit. It also used AD 712 devices as current buffers for tape outputs. A lot of high end designers thought real highly of it. You might give the article a look. Its in issue 1/94. If you cant find it let me know and I will scan it for you.
 
Scans of the Valkyrie article are available in the files section at <a href="http://groups.yahoo.com/group/tnt-audioaddicts/">tnt-audioaddicts</a> you will have to join the group to see the files

You might be interested in Panasonics 4 terminal <a href="http://www.flint.co.uk/product/pdfs/FP1ECG.PDF">SP Caps (.pdf)</a> for PSU bypass, looks good on paper.

<center>Regards
James</center>
 
I tried the Valkyrie with the AudioXpress PCB and it oscillated like crazy. Didn't have a scope of my own at the time, so I gave up. Been thinking of resurrecting it as a headphone amp, now that I have a scope.
I don't quite understand why you need a 100 MHZ op amp for audio. I think you'll be causing yourself more grief than anything. If you don't have experience in RF circuitry layout, steer clear of it.
 
mlloyd1 said:
Worked like a charm; sounded great, very stable
That's encouraging, I'll have to fire mine up again. Sorry I don't remember the details, it's been a few years. Now that I have a scope and resources like this site, I should have a much better chance of finding and fixing the problem. I'm thinking it would make a good headphone amp.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.