Dr. Mabuse reports running into problems when turning the cutoff offset
pot all the way in which the circuit was drawing excessive current. I
was unable to reproduce this problem using either the fancy lab supplies
at work or the MOTM power supply (essentially a Power One with a fancy panel)
I have at home. Dr. Mabuse
uses a homebrew barebones power supply that can generate up to 2 amps and
that doesn’t have any particular current limiting circuitry in it.
He writes (paraphrasing): “I burnt-up pot after pot. I tried ‘simple’
V-regulation schemes. I tried a couple of diode drops from the 15V rail…
burned up the pot! I tried a few transistor-based regulation schemes
and I burned up the transistors. I
ried a few combinations of the above and I
burned up the transistors AND the pot!!!!”
So Dr. Mabuse built a variation of the full +13.5 V supply used in the
original Easel. He notes that it is also a tightly controlled current
limiter, and once he built it, he got his LPG to behave and to stop
burning up components. On the original Easel, the supply is on Board 3:
see here. Doc used a 741 intead of a 4136 (I’ve been using
4558s in place of 4136s in all my PCBs), a 2N3906 in place of the 2N2905
for Q2 (although he later used a 2N2907, which can source more current
than a 2N3906). He also used an 8.2K resistor in place of the 10K 1% for R31.
Doc did some more experiments. He wrote:
I modeled it on
EW many ways…including: just like I had it wired on the bench …
and… (as you suggest)
with no voltage divider – right off the 15 Vdc rail –
… and in that case, it sucked 1.63ma through that pot as politely as
you please.
And in all the other cases, EW predicted even lower current draws.
So I tried wiring it up in the real-world again.
Since my power supply is so ‘raw’ i put a
1K 1watt resistor in series with the 13.5 v input
point and measured the current going in.
Everything behaved pretty close to the modeled
results until about the last 5% of the pots rotation…
then the current shot up precipitously
to about 350ma. This happened right about 14.2 vdc at the wiper.
When I tried using a single transistor as a ‘poor-mans’ V-reg (at 13.5v), this current spike
occurred only when the pot was floored but
it cooked the transistor slowly… over time.
If I tried just dropping the 13.5 source down from 15 Vdc
with a couple of diodes (4001’s) the spike occurred
at the same point on the pot but it seemed
to break the diodes down suddenly, and that cooked the pot.
But when I built Don’s little opamp+PNP supply,
the current limiting seems to tame
this aberration and the circuit behaves very nicely.
I’m out of my depth to try and find the root cause and I’ve
contentedly built the Buchla design for the 13.5 ‘current-governor’ supply.
Grant collaborated with me a little and he
mentioned something intriguing that might be worth a look. I
note that you put in the precautionary resistors around the 4016 switch.
Be that as it may, Grant mentioned that that not all the old 1st-gen CMOS
(like 4016s & 4001s) were created equal. Some of the old ones lacked
internal diode protection that was added on later revisions and
this sudden mysterious greedy hogging of current could be an
old-style CMOS quirk.
This is purely a gut feel… but it looks to me like something
is hitting a systemic voltage threshold and going all superconductor.
I’ve seen ‘sort of’ similar conditions before and all
the previous similarities had CMOS involved. Don probably never
noted it because, like you, he was using real-deal
domesticated multi-output power supplies rather than the bareback DIY
stuff I use.
I really have no clue what might be going on here. As I said, I have
not been able to reproduce this issue, and looking over the circuit,
I can’t figure out where the crazy current draw might be coming from,
even if I try to imagine the CMOS getting wacky.
I am very interested to hear about people’s experiences with this.