2N7000 FET in sawtooth VCO

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!
Post Reply
User avatar
Bergfotron
Common Wiggler
Posts: 159
Joined: Tue Aug 23, 2011 8:24 am
Location: Sweden

2N7000 FET in sawtooth VCO

Post by Bergfotron »

I just tested to replace the 2n4391 with a 2N7000 in two of my VCOs.
The Minimoog style VCO didn't oscillate at all.
The Basic VCO (ARP 4027-1 clone) did oscillate but the sawtooth waveform was not correct.
So, unfortunaltely, the 2N7000 wil not work as a direct replacement in those VCOs. It is possible that it might work if some other changes are made to the circuits. But as this can affect tuning stability, tracking etc, it will require substantial development work.
frijitz
Super Deluxe Wiggler
Posts: 1679
Joined: Mon Aug 31, 2009 3:37 pm
Location: NM
Contact:

Re: 2N7000 FET in sawtooth VCO

Post by frijitz »

Bergfotron wrote:I just tested to replace the 2n4391 with a 2N7000 in two of my VCOs.
The Minimoog style VCO didn't oscillate at all.
The Basic VCO (ARP 4027-1 clone) did oscillate but the sawtooth waveform was not correct.
So, unfortunaltely, the 2N7000 wil not work as a direct replacement in those VCOs. It is possible that it might work if some other changes are made to the circuits. But as this can affect tuning stability, tracking etc, it will require substantial development work.
I spent a lot of time looking at MOSFET-discharge Saw oscillators. There is a good circuit posted at e-mus. An enhancement mode device cannot generally replace a JFET because the biasing requirements are different. Also the 2N7000 has a huge input capacitance that gives gigantic switching transients. I also used MOSFETs for my TZVCO:
http://home.comcast.net/~ijfritz/board% ... _a_doc.pdf

:grin:

Ian
User avatar
Bergfotron
Common Wiggler
Posts: 159
Joined: Tue Aug 23, 2011 8:24 am
Location: Sweden

Post by Bergfotron »

Thanks for your reply Ian (and nice to see that you're here too). This thread was in response to somebody who claimed that the 2N7000 could indeed be used. So I just thought I'd give it a try.
As you point out, the biasing requirements are different. But if the FET is driven by a comparator that swings rail to rail, that doesn't have to be a problem (in theory).
frijitz
Super Deluxe Wiggler
Posts: 1679
Joined: Mon Aug 31, 2009 3:37 pm
Location: NM
Contact:

Post by frijitz »

Bergfotron wrote:Thanks for your reply Ian (and nice to see that you're here too).
Good to see you're around. Seems like it's been a while?
This thread was in response to somebody who claimed that the 2N7000 could indeed be used.
Yes, and I was very surprised by that post, given the author's reputation.
But if the FET is driven by a comparator that swings rail to rail, that doesn't have to be a problem (in theory).
Generally that's probably true -- you just have to be sure the pinch-off requirements are met. And you have to be aware of the protection diode and what it can do to you.

:grin:

Ian
paults
mice nuts
Posts: 4650
Joined: Fri Jan 16, 2009 12:46 am

Post by paults »

I *never* implied it was a 'drop in'. Sheesh.......I said it would work in a VCO circuit, just didn't say which one or if you had to actually do a *bit* of work.

FWIW my VCOs use 'reverse alpha' bipolar transistors and they do require about 30min of calculations to get the biasing correct. But they work great and no supplier issues.
User avatar
Dr. Sketch-n-Etch
Super Deluxe Wiggler
Posts: 8658
Joined: Thu Jan 21, 2010 1:46 pm
Location: Vancouver, BC, Canada

Post by Dr. Sketch-n-Etch »

I still like the 2N5485 JFET. 2N4391 is great, but 2N5485 is almost as good in VCO cores, better in other applications (like saw shapers), and is readily available in both through-hole and SMD (MMBY5485). When you're in production, it's nice to have parts that can fulfull multiple roles.

My $0.02

I also really like Ian's approach of just using an analog switch (although they are dearer and take up a bit of space on the board).
We've moved fast. We need love. A part we offer is our only freedom.
frijitz
Super Deluxe Wiggler
Posts: 1679
Joined: Mon Aug 31, 2009 3:37 pm
Location: NM
Contact:

Post by frijitz »

Here is my MOSFET-discharge Saw VCO and its switching behavior. The switching pulse is shaped to optimize the discharge.

Image

Image

I wasn't able to do anywhere nearly as well with a 2N7000.

:grin:

Ian
User avatar
gasboss775
Super Deluxe Wiggler
Posts: 1212
Joined: Tue Feb 02, 2016 10:34 am
Location: Scotland

Post by gasboss775 »

The Drain and Source pins are swapped between the VN0104 & the 2N7000 the gate is in the middle in both cases. If this hadn't been taken into account it could explain the poor performance.
MEEEEEEOOOOOOWWWWWWW!!!!!!!!
homeless_peep
Common Wiggler
Posts: 72
Joined: Thu Nov 17, 2016 7:55 am
Location: greece

Post by homeless_peep »

A key spec for a discharge transistor on a sawtooth oscillator is it's On resistance. When the transistor turns on, the capacitor will discharge at a rate proportional to the resistance it sees towards the low/high node (ie, Vref or ground). As far as speed is concerned, a slower transistor will need more time to turn on, thus the sawtooth will be left to increase/decrease a little bit further than the threshold set by the comparator/trigger. This will increase it's amplitude, while causing a decrease in frequency, as you go up.
2n7000 has an on resistance of 1.5Ohm, although some models show up as 5Ohms.
I think a reasonably fast power transistor is just ideal for this kind of job. They are fairly cheap too, and feature resistances much lower than 200mOhm.
frijitz
Super Deluxe Wiggler
Posts: 1679
Joined: Mon Aug 31, 2009 3:37 pm
Location: NM
Contact:

Post by frijitz »

homeless_peep wrote:A key spec for a discharge transistor on a sawtooth oscillator is it's On resistance. When the transistor turns on, the capacitor will discharge at a rate proportional to the resistance it sees towards the low/high node (ie, Vref or ground). As far as speed is concerned, a slower transistor will need more time to turn on, thus the sawtooth will be left to increase/decrease a little bit further than the threshold set by the comparator/trigger. This will increase it's amplitude, while causing a decrease in frequency, as you go up.
2n7000 has an on resistance of 1.5Ohm, although some models show up as 5Ohms.
I think a reasonably fast power transistor is just ideal for this kind of job. They are fairly cheap too, and feature resistances much lower than 200mOhm.
Two problems with these ideas, as has been discussed previously. First, power transistors have large internal capacitance, which leads to problems with charge storage (pedestal voltage). Second, there is no need to have the discharge rate faster than the integrator can follow. Too fast a discharge leads to problems with large voltage spikes. This is why my design has a pulse shaping network at the gate.

But if you come up with a better design, please show it to us.

Ian
User avatar
okelk
Wiggling with Experience
Posts: 471
Joined: Mon Jan 13, 2014 4:16 am
Location: Vienna, Austria

Post by okelk »

frijitz wrote:This is why my design has a pulse shaping network at the gate.
That's interesting, but I don't quite understand it.
Would you care to explain why a different shape of the discharge-pulse helps?
I always thought it's just supposed to go up as quick as possible...

By the way....the vco that's just beiing discussed at this thead:
1011 VCO
seems to use a 2n7000 in it's saw core...

I personally only used a 2n7000 as a switch in a triangle to saw converter (classic optional inverter configuration)...It worked fine..but I dind't spend any time trying to optimize it in any way....it was just good enough at the first try...
User avatar
SlightlyNasty
Common Wiggler
Posts: 221
Joined: Wed Nov 25, 2015 9:18 pm
Location: Adelaide, Australia

Post by SlightlyNasty »

okelk wrote:By the way....the vco that's just beiing discussed at this thead:
1011 VCO
seems to use a 2n7000 in it's saw core...
Yes, it all depends on the degree of performance you're after. The 1011 is supposed to be a little rough and characterful, so weird switching transients weren't a huge concern. I wasn't afraid to just shape and filter them out with RC networks where necessary. Part of the fun of that design is seeing what can be squeezed out of common jellybean parts.
Post Reply

Return to “Music Tech DIY”