Discrete Opamp Open Design

Bravo! I haven't looked at the layout yet, but this is a great start.
Thanks for the encouragement! Since there was a lot of comparison to the JE990 a hundred pages or so back, I thought there would be a lot of support for a module that at least came close to that form factor. As you can tell from previous comments, I guessed wrong.

1) Perhaps if dual-transistor BJTs in SOT23-6, SOT363 or SMD6 were used, the outline could be shrunk to the API-2520 form-factor.
Yes, those substitutions would save acreage. Also going to 0603 resistors. But based on others' previous posts these would NOT be popular actions. (The dual transistor packages in particular reduce the constellation of potential transistors from dozens or hundreds, to about a dozen.)

If I changed the output device patterns to TO126/TO225 padstacks rather than the combination SOT23 and SOT223, dropped support for thru-hole capacitors, and used dual-transistor packages, I might be able to get it down to the API-2520 footprint size.

2) Similarly, dual-sided SMD layout with the gain stages on one side, and the diamond buffer on the other side, could also help shrink the layout.
I frown on mounting components on both sides of the PWB in this application. The bottom-side parts are likely to get squashed, or shorted, against a chassis when the module is installed in a chassis. Using contact pins with a longer flange (I said LONGER, not wider) such as Mill-Max 3125-2-00-00-00-00-08 ($0.22 ea from Mouser in 10's) is a viable work-around but not my preferred approach.

I'd be more inclined to make a two-story assembly, as Scott showed in his last prototype. Even that's not a good solution for physically small assemblies like this, since the connections from one PWB to the next use up a significant portion of the available acreage. (I.e., adding a second PWB to the assembly doesn't double the available board space - after subtracting the area used by the interconnections, the increase may be only 50% .)

3) Apart from the Mill-max gold-plate swage pins suggested elsewhere, Vector K31C/M tin-plated pins of 0.042" diameter, may be a lower-cost and/or more easily available alternative.
The parts list suggests Mill-Max p/n 3103-2-00-15-00-00-08-0 ($0.17 ea from Digi-Key in 25's) or Keystone 1405-2 ($0.49 ea from Mouser in singles).

Dale
 
. . . but here are some of my thoughts ...
Thanks for taking the time to evaluate this. I'm taking your suggestions seriously. For now, many of your comments can be incorporated by simply leaving parts off the assembly, though eliminating their locations on the PWB is attractive. The signal input traces are quite short (conscious effort to make them short), but I can open up the bottom-side ground plane in that vicinity.

Your R27 source resistor will be VERY twitchy. Scott recommended tweaking R9 instead and sims confirm this is less critical.
Can you suggest a suitable adjust range? Given that the nominal value is 250 ohms, should I try for a range of 200 ohms to 300 ohms, or 100 to 600 ohms?

Dale
 
"Depending on your experience with SMT assemblies, and the soldering tools available to you, removal and replacement of individual components will be slightly, to extremely, challenging." dscisholm

TO SAY THE LEAST! I think Scott's original idea of a sandwich of two boards seperated by spacers is the way to go for thru hole. If both boards have ground planes then arranging them face to face with the components on the ground plane side will provide an excellent shield for the VHF jfet input devices. That's what I'm gona do. Also the ground plane side will still allow a hybrid of both SMT and thru hole parts allowing both SMT and thru hole devices. The only drawback is once the two boards are together it will be a real downer to replace parts. So it would be necessary to pre-test the input board and output board before they are joined or place the parts on the two outside sides. Also to92 dif-amp jefts and current mirror BJTs allow their flat sides to be epoxed together making them more temp stable to drifting and also pushing them closer to the board surface. The board just needs more room to allow easier soldering for home constructors. Just my two cents! Ray
 
... Since there was a lot of comparison to the JE990 a hundred pages or so back, I thought there would be a lot of support for a module that at least came close to that form factor. As you can tell from previous comments, I guessed wrong.
...
I'd be more inclined to make a two-story assembly, as Scott showed in his last prototype. Even that's not a good solution for physically small assemblies like this, since the connections from one PWB to the next use up a significant portion of the available acreage. (I.e., adding a second PWB to the assembly doesn't double the available board space - after subtracting the area used by the interconnections, the increase may be only 50% .)

It would seem that DIP8 would be more popular for most DIYers, and API2520 for those who maintain/own professional studio recording gear.

DIP8 duals are out of the question for SWOPA, but singles are probably at the fringe of feasibility with a two-story assembly with the gain stages on one, and the buffer on the other. Both boards can be mounted perpendicularly on the base board, which will have the pads for the DIP8 pins, and maybe a few odds and ends like bypass caps, compensation networks, etc. Signals may be routed between the gain stage and buffer stage through struts (of which at least 2 are needed anyway for mechanical support), or to rectangular pads at the lower edge, to the base board and across to the other side (of course, this is tricky because of parasitics, crosstalk, etc., but it should be ok for nodes without significant voltage swing, or tracks without significant current swing).

Up to 8-12 rectangular pads are feasible at the board edge, though it's more common to use around 5-6. Some of these will be consumed by V+, V-, Comp, Earth, In-, In+, etc., but a few may still be feasible for inter-board signals.
 
There are a host of footprints out there; The Deane Jensen 990, The Pacific Recorders 990, the John Hardy 990, the API2520 and DIP8. And what about FM Acoustics modules? I think there probably needs to be an adapter for each case. And in each case I don't think this is gona be plug and play. Ray
 
diyAudio Member RIP
Joined 2005
Is there a consensus that Susumu "RG" series thin-film resistors are suitable in this design? They're about twice the cost of the common thick-film parts, and would add about US$2.00 to the BOM cost.

Or should I be trying to find some other construction, such as a SMT metal film resistor?

Dale
I've had good results with them, and they seem readily available. They are not needed everywhere, and they are costly, and still there is a small size effect it appears (bigger is better, see the ligo material and others).

But as Scott pointed out, it's not an SMD issue per se, and he once cited an example of very thin on-chip parts (800 atomic layers if memory serves) that work just fine. Evidently that's enough layers to make spooky mesoscopic effects still remain small ;)

Brad
 
Scott,

What is the recommended offset nulling scheme?

WRT the schematic of the input gain stage in your post #2643, a trimmer between the upper current mirror emitters (U1 emitter and W6 terminal) with the wiper connected to positive rail?

Also, is output current limiting necessary / advisable for the output buffer?

WRT the schematic in post #2593 a diode clamp (two anti-paralleled connected diodes) between output and the center tap of the 30K9 resistor to be split into two resistors?

Thank you (for the great learning experience)

Giorgio
 
Scott,

What is the recommended offset nulling scheme?

WRT the schematic of the input gain stage in your post #2643, a trimmer between the upper current mirror emitters (U1 emitter and W6 terminal) with the wiper connected to positive rail?

Also, is output current limiting necessary / advisable for the output buffer?

WRT the schematic in post #2593 a diode clamp (two anti-paralleled connected diodes) between output and the center tap of the 30K9 resistor to be split into two resistors?

Thank you (for the great learning experience)

Giorgio

Yes, that offset trim is prefered. The diode clamp should go from the output to the input of the output stage not the 30K9 resistor. That should be enough protection for normal use.
 
Scott,

Within the ad797 datasheet there is a reference to using an output resistor for almost any capacitive load.
I dont fully understand its purpose.
Can you explain?
And would it be worth while to include such a device on the circuit board?


This is common, the output impedance rises with frequency so it looks inductive and a cap load can make it oscillate. Usually a small R like 25-50 Ohms helps. The addition of a capacitor from the output back to the inverting input provides AC feedback without "seeing" the cap load. This stuff usually helps well outside the audio band.

BTW even an open-loop emitter follower can oscillate, but that is usually 100's of MHz and not seen easily on the scopes DIY'ers often use.
 
Disabled Account
Joined 2012
Re making a th circuit --- sm for the input transistors is OK and the rest th. Size isnt important... meaning it doesnt have to be the absolute smallest possible size. Use "standard" part size. That way most brands will fit. Is a kit being made of pcb and parts? Or better, yet assembled pcb? If not, it doesnt matter what is in the BOM.... Susumu or otherwise.

-RNM
 
Link to a prototype folded Kaneda-style Class-A DIP8 single opamp module, approx. 11mm x 14mm in size:

http://www.diyaudio.com/forums/analog-line-level/10243-kaneda-preamp-11.html#post3348288

The resistors are 0603 size; all actives are SOT23, except a Rohm dual-PNP in SMT6; and the rail-to-rail bypass is a Panasonic ECHU PPS film-foil in 1210 size, mounted on the underside. All signal traces are on the component side only, allowing the possibility of a single-sided board in future revisions.

I cheated on the prototypes shown, by using a Zetex ZRA125 1.25V bandgap reference instead of the BAV99 dual-diode in the schematic - I just wanted to be sure that the first prototypes would work, i.e. produce some sound, which they did on the very first test.
 
Thin film resistors consist of a metallic deposition (made by vacuum deposition or sputtering process) with a 50 to 250 angstrom thickness on a ceramic substrate [recall that one angstrom (Å) = one nanometer = 10-10 meter].

I can spot that as wrong 15 min after 3 drinks with dinner, search doesn't find CVD in the article either - are they sure they're up to date on R film tech?
 
Last edited:
A little more insight
thin film resistors versus thick film resistors how they compare
However, the heat generated during laser trimming causes micro-cracks on a thick film resistor and therefore affects the short term and long term stability. Thin film resistors need less laser power than that needed for thick film resistor. Thin film resistors do not exhibit micro-cracking during laser trimming. Consequently thin film resistors show superior stability and noise performance.
 
This is common, the output impedance rises with frequency so it looks inductive and a cap load can make it oscillate. Usually a small R like 25-50 Ohms helps. The addition of a capacitor from the output back to the inverting input provides AC feedback without "seeing" the cap load. This stuff usually helps well outside the audio band.

BTW even an open-loop emitter follower can oscillate, but that is usually 100's of MHz and not seen easily on the scopes DIY'ers often use.

Hi Scott,

The information Ive seen showed improvement for extremely low distortion levels, like around -100db.
Is that caused by oscillations?
 
Transistor selection

I'd like to go full SMD to avoid the somewhat complex assembly of TO92 devices in the output stage. OTOH, I'd like to retain the thermal tracking feature of the PTH configuration while driving low loads.

I've been looking for dual NPN/PNP transistors in SMD packages capable of some power dissipation and the only parts that I came up with are the NXP PBSS4041SPN:

http://www.nxp.com/documents/data_sheet/PBSS4041SPN.pdf

It seems to me that those parts are designed for (static) switching rather than linear operation. They are also comparatively slower than 2N4401 / 2N4403, for instance.

Questions for the Semiconductors Gurus:

- are those parts any suitable for the output diamond buffer?
- any similar, better, parts?

Thank you.

Giorgio