EUVL said:Steen,
If you have an application to test, and promise to publish the results, send me a private mail (you have the address).
Patrick
YGM.
Hi EUVL. Those little op amps look awesome! If you need more pcb real estate you could try stacking two or more boards. You could also use different SOT23-6 transistors that have matched devices in a SOT23 body with 6 pins to save some space and get better thermal tracking.
BC847BS/557BS is quite nice and dirt cheap.BWRX said:Hi EUVL. Those little op amps look awesome! If you need more pcb real estate you could try stacking two or more boards. You could also use different SOT23-6 transistors that have matched devices in a SOT23 body with 6 pins to save some space and get better thermal tracking.
Attachments
> You could also use different SOT23-6 transistors that have matched devices in a SOT23 body with 6 pins to save some space and get better thermal tracking.
I know. I mentioned in my design description that I started off with SOT-23-6. But I want to use JFETs and there is no matched JFETs in SOT-23-6 that has > 25mS transcoductance, Idss around 10mA and low noise. And the current layout is actually better than SOT-23-6, believe me.
As for thermal tracking, if you use the right potting compound, it is not a real problem. As it is, the biggest thermal drift is because of the different Temp Coef between the BF86x's and the ZVP3306, and not the JFET themselves. All the high dissipation components are glued to the heatsink directly using Arctic Silver. And I was amazed how effective it was.
I am not using bipolar, as they require current mirrors, biasing diodes, etc. A current source is much easier to make from JFETs. Bipolar means more components, which in turn means more PCB real estate.
> Your competitors have huge designs compared to yours.
I know. I did an extensive survey. I thought about posting some of those pictures (like Forssell, API, Burson, Melcor, JE-990, .....), but then there are too many copyright issues. I think people who knows the topic, like yourself, knows what they are looking at.
In fact, if I were to make a couple of thousands of these, I have a even better means -- buy LSK389 bare dies from Linear Systems and probably Zetex and do direct chip on board with wire bonding. It will then be packaged like a normal DIP-8. It is handy to be working in semiconductors.
![Wink ;) ;)](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)
> BC847BS/557BS is quite nice and dirt cheap.
I want to see how you match them.![Smile :) :)](data:image/gif;base64,R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7)
I am too old to solder anything much smaller than SOT-23 these days.
Patrick
I know. I mentioned in my design description that I started off with SOT-23-6. But I want to use JFETs and there is no matched JFETs in SOT-23-6 that has > 25mS transcoductance, Idss around 10mA and low noise. And the current layout is actually better than SOT-23-6, believe me.
As for thermal tracking, if you use the right potting compound, it is not a real problem. As it is, the biggest thermal drift is because of the different Temp Coef between the BF86x's and the ZVP3306, and not the JFET themselves. All the high dissipation components are glued to the heatsink directly using Arctic Silver. And I was amazed how effective it was.
I am not using bipolar, as they require current mirrors, biasing diodes, etc. A current source is much easier to make from JFETs. Bipolar means more components, which in turn means more PCB real estate.
> Your competitors have huge designs compared to yours.
I know. I did an extensive survey. I thought about posting some of those pictures (like Forssell, API, Burson, Melcor, JE-990, .....), but then there are too many copyright issues. I think people who knows the topic, like yourself, knows what they are looking at.
In fact, if I were to make a couple of thousands of these, I have a even better means -- buy LSK389 bare dies from Linear Systems and probably Zetex and do direct chip on board with wire bonding. It will then be packaged like a normal DIP-8. It is handy to be working in semiconductors.
> BC847BS/557BS is quite nice and dirt cheap.
I want to see how you match them.
I am too old to solder anything much smaller than SOT-23 these days.
Patrick
EUVL said:> ... I have a even better means -- buy LSK389 bare dies from Linear Systems and probably Zetex ...
Patrick
Do these mfg'ers sell bump die (aka flip chips)?
Of course the LSK389 is only available in SOT-23-6 in theory.
You could try this one from NXP:
PMBFJ620 dual-JFET SOT363 (RS components P/N 626-3320).
We have done some more simulation and testing. Basically you can get stability for almost any gain other than +1 (unity gain buffer). But then for unity gain buffer, one might just as well use a pair of BF862 without any feedback.
For gain of -1 (unity gain inverting), we found that it is best not to use the compensation cap but just a cap in parallel the feedback resistor. Bandwidth is never a problem, always way above 200kHz.
We plan to test this in a Buffer-Filter-Buffer phono preamp with single stage gain of 100. Will take a while. Need to make and trim 4 new ones.
Patrick
For gain of -1 (unity gain inverting), we found that it is best not to use the compensation cap but just a cap in parallel the feedback resistor. Bandwidth is never a problem, always way above 200kHz.
We plan to test this in a Buffer-Filter-Buffer phono preamp with single stage gain of 100. Will take a while. Need to make and trim 4 new ones.
Patrick
EUVL said:One more ....
I love the shape of top layer - reminds me of '70s hand-drawn PCBs, just inverted
Nice work, BTW
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Pass Labs
- Pass Discrete Opamp in DIP-8 Package