Discrete Opamp Open Design

[scott wurcer]

I'll get around to the explanation, promise.

A thought occured to me last night, there are several build to print houses in Taiwan that offer some surprising value. With Stinious' BOM down in the $1 range in even modest volume the need to solder all those tiny packages could be eliminated.

Another fine point, the resistors along the rails are in a 2,1,3 ratio so 6 resistors of the same (~125 Ohm) value off of the same reel would eliminate having to only get close on the ratios. I would use either 124 or 127 Ohms. It adds 3 resistors but eliminates 2 values. In any case the 375 Ohm ones are not critical so doing this at the input only adds one resistor.

[qusp]

thats an interesting idea mate, but I actually think its a good opportunity for some of these guys that are seasoned veterans otherwise, but are a bit scared of SMD to get to know the new packages they will have to be dealing with, on a small scale project with not many parts, easy to troubleshoot and not much of a loss if it goes up in smoke. some of course are totally comfortable with it, but some could use a push IMO.

I mean no disrespect here, many are much more accomplished DIYers than I am, just not so comfortable with SMD and these are small, but really not very difficult parts to solder

[qusp]

more of a worry than soldering is matching the little suckers

[grhughes]

BC846BPN ?? How well matched are these complementary devices? They are obviously on the same substrate. Probably more matched that trying to match them ourselves??? Also the resistors around the diff amp: How well matched should those be: .01, .001?? Also I would like to know which transistors need to be thermally coupled to the output transistors. It seems this has been redrawn to suit each contributor and I'd like to stick to Scott's latest schematic. I know that diff amps and constant current sources and sinks if they are current mirrors need to be matched and thermally coupled to not drift. Obviously this simulation will need to be built and go through a parts substitution phase before it is safe to order parts? Waiting in the wings. Very enthused and interested! Ray

[bcarso]

Originally Posted by grhughes

BC846BPN ?? How well matched are these complementary devices? They are obviously on the same substrate. Probably more matched that trying to match them ourselves??? Also the resistors around the diff amp: How well matched should those be: .01, .001?? Also I would like to know which transistors need to be thermally coupled to the output transistors. It seems this has been redrawn to suit each contributor and I'd like to stick to Scott's latest schematic. I know that diff amps and constant current sources and sinks if they are current mirrors need to be matched and thermally coupled to not drift. Obviously this simulation will need to be built and go through a parts substitution phase before it is safe to order parts? Waiting in the wings. Very enthused and interested! Ray

http://www.nxp.com/documents/data_sheet/BC846BPN.pdf

They appear to have done an admirable job of matching current gains and base-emitter voltages in the vicinity of 2mA Ic. However, the statement early on in the datasheet, "no mutual interference between the transistors", suggests that they are NOT on the same substrate.

[scott wurcer]

Originally Posted by bcarso

http://www.nxp.com/documents/data_sheet/BC846BPN.pdf

They appear to have done an admirable job of matching current gains and base-emitter voltages in the vicinity of 2mA Ic. However, the statement early on in the datasheet, "no mutual interference between the transistors", suggests that they are NOT on the same substrate.

Unless it's SOI. Note there are no side to side specs. I think I will order some and give a look, if anyone has some a fine sandpaper disk in a Dremel should be enough to see if there are one or two die.

[PHEONIX]

Hello Scott,

I feel like I'm asking a stupid question but the input stage of the discrete opamp is a pair of BF862 don't these need to be matched for optimum performance. How do you get matched parts , I suspect matching these smd parts is not so simple and a lot of work. If this aspect has been covered in this thread could somebody point me to it.

Arthur

[bcarso]

Originally Posted by PHEONIX

Hello Scott,

I feel like I'm asking a stupid question but the input stage of the discrete opamp is a pair of BF862 don't these need to be matched for optimum performance. How do you get matched parts , I suspect matching these smd parts is not so simple and a lot of work. If this aspect has been covered in this thread could somebody point me to it.

Arthur

Unfortunately, test sockets for SOT-23 devices are expensive. I recently bought one for about 35 dollars iirc and specifically to match parts like the BF862. Ideally, sort them for Idss and threshold voltage, and make sure to put some damping R in series with the gate lead connection to prevent high frequency oscillations.

But testing and sorting is tedious, no doubt about it.

[rsavas]

The test fixture can be as simple as some wave-solder type pads and a bit of weight on the component, to make the contact (Even tweezers would work).. I know it is a bit delicate, but it is possible without being fancy. SOT23, piece of cake, boy I hate DFN, damn cell phone people :-)

[scott wurcer]

Originally Posted by PHEONIX

Hello Scott,

I feel like I'm asking a stupid question but the input stage of the discrete opamp is a pair of BF862 don't these need to be matched for optimum performance. How do you get matched parts , I suspect matching these smd parts is not so simple and a lot of work. If this aspect has been covered in this thread could somebody point me to it.

Arthur

The LSK389 works almost as well for those who can't bother. I took a PC board with some pads at SOT23 spacing and set it up to measure Idss. Using a plastic tweezers I pressed the parts down one at a time, you can do almost a whole tube in less than an hour. The plastic is both non-conductive and prevents finger heating.