Fun with JFETs and servos
- Thread starter jubal81
- Start date
Chuck D. Bones
Circuit Wizard
Might be me, maybe I'm a little %$&^ed-up the head, but those integrating capacitors should be between the output and the -in terminals of the opamps. Also try CA3130 if you want something that can swing to the rails. R5 and R12 could be larger, much larger if need be. Gives R6 and R14 more leverage. The technical term is control authority.
jubal81
Well-known member
The more I think about it, the more I think the whole idea is a little %$&^ed-up in the head. Feels like I'm on Mouser shopping for the right $200K supercar someone can leave parked in their garage just in case their 1987 Chevy Corsica needs a jump start.
In practicality, I think it could be an option if you're Brian Wampler and building 10,000 of these and the extra $ on parts costs less than having someone manually sort or bias JFETs. For DIY, you can get trimmer pots at Tayda for 17 cents and it takes just a few seconds to set and forget a bias voltage on the JFET gate.
I'm gonna keep tinkering with it when my parts arrive, though, just because it's an interesting puzzle to solve and could be useful for something down the road. I see what you mean on R5 and R12. Always seemed kinda wonky to me, but I was going with that diagram from Elliot Sound. I'll also experiment with those caps. I forgot those on the first go and the system wasn't working.
BTW, you think a TLC085 would be an option? 4 channels for $4.
Chuck D. Bones
Circuit Wizard
I don't think that auto-bias will totally eliminate the need to screen JFETs for production builds. When a JFET has a 6:1 Vp range (ex. 2N5457), there is no way around having to select devices that fit into a narrower range. There is an economy of scale when screening parts. If someone buys 10,000 JFETs, they can pay a couple $K to have them all run thru an automatic tester and sorted, either by the manufacturer, distributor or a 3rd party testing house. It's very common in the electronics manufacturing industry.
The mu amp circuit mostly eliminates the need to screen JFETs for Vp and Idss because the top FET sets the drain voltage and the bottom FET sets the drain current. FETs at the extreme ends of spec range can still cause problems, but no trimmers or select-in-test resistors are required.
My opinion: auto-biasing JFETs is the tail wagging the dog because JFETs don't drift. Set the bias once by selecting a resistor in test and you're done. Still, there is value in experimentation so don't let my opinions dissuade you.
To quote Friedrich Nietzsche:
“It is hard enough to remember my opinions, without also remembering my reasons for them!”
The mu amp circuit mostly eliminates the need to screen JFETs for Vp and Idss because the top FET sets the drain voltage and the bottom FET sets the drain current. FETs at the extreme ends of spec range can still cause problems, but no trimmers or select-in-test resistors are required.
My opinion: auto-biasing JFETs is the tail wagging the dog because JFETs don't drift. Set the bias once by selecting a resistor in test and you're done. Still, there is value in experimentation so don't let my opinions dissuade you.
To quote Friedrich Nietzsche:
“It is hard enough to remember my opinions, without also remembering my reasons for them!”
Chuck D. Bones
Circuit Wizard
Those will work. If you need four channels, then it's cheaper than buying four CA3130's. For that circuit, we want an opamp that meets these criteria:
- very low bias input current -> FET input
- output can swing close to the -rail
Chuck D. Bones
Circuit Wizard
Q: What happens if you need to bias the gates more negative?
A: The source resistor gets bigger, which consumes headroom.
I think you're better off making the source resistor a trimpot and bypassing that as req'd.
jubal81
Well-known member
Sweet! That sure simplifies things. I've always been able to gate bias, so I guess I hadn't considered that problem.
Chuck D. Bones
Circuit Wizard
I feel like we've come full-circle.
