the Tail of a Long Tail (transistors)

[JohnRoberts]

John, Peavey stories are always welcome, you can drop the TMI brackets

Thanz.. I learned a lot during my 15 years there. I thought this project could be an opportunity to learn about full custom IC development. Back in the 70s I got to work with Monochip Semi-custom ICs. They used custom metallization layers to configure standard IC substrates loaded with multiple uncommitted semi devices. Instead I ended up trying to teach the IC engineer how ICs work inside. It became clear I was talking to the wrong guy inside his company but it dawned on me, that including an OTA that didn't suck inside a full custom IC would be more than a simple cut and paste exercise. Since real chip makers struggle with OTA performance unlikely to be in a generic library. While the theory about how OTA s work is not that difficult, in practice not trivial.

JR

 

[abbey road d enfer]

Since real chip makers struggle with OTA performance unlikely to be in a generic library. While the theory about how OTA s work is not that difficult, in practice not trivial.

Testimony is that nothing has really improved since the 13700 18 years ago. Improving performance by an order of magnitude would require running the core at 100 mA doesn't look like a viable proposition neither running 100 chips in parallels, when a Blackmer/Buff achieves two orders of magnitude improvement with about 2mA in the core.

 

[abbey road d enfer]

@ Analog_Fan
BTW, it's tale of..., if I catch the gist of it.

 

[JohnRoberts]

Testimony is that nothing has really improved since the 13700 18 years ago. Improving performance by an order of magnitude would require running the core at 100 mA doesn't look like a viable proposition neither running 100 chips in parallels, when a Blackmer/Buff achieves two orders of magnitude improvement with about 2mA in the core.

Yup, that's about when the junior IC engineer's eyes glazed over, as I tried to explain that a well executed OTA would require committing a bunch of silicon at making some low Rbb devices for the OTA input stage LTP.... He tried to tell me that it was no problem just add some degeneration resistors in series with the emitters of small geometry transistors. Real estate is the major cost driver for full custom ICs. I suspect an approach similar to LM394 with a bunch of small geometry devices in parallel might work... but we never made it to that stage.

My cost doing it with separate off the shelf parts was not much more than $1 (while the OTA we were using were a house number selected by the mfr for low control feed through.)

JR