Replies to Minipulse Plus

[Davor]

Hi Davor, no drama, its all good,
Sorry if you answered me somewhere else but I missed that reply
my questions where more a response to silver dollar's reply #325, sorry for confusion .

I built the time base but could not get it to work properly so pulled it apart, but on retesting the 4017
it tests faulty, and does not cycle correctly, so will build the time base again , thanks for the reply cheers.

In that case just hold a little bit longer as I have a small update on the way, and I'll see to make it myself. I replaced the Schmitt NANDS with a more common CD4011. I realised that I have a spare gate that calls for making a more controllable Schmitt from a pair of them. A small problem with CD4093 is that the hysteresis voltage is not directly proportional with the supply voltage, and you can't know how large it actually is for a given chip.

[CAS]

What version board do you have?.
You mean pin 7 on U2? to Q2 collector
that is the sync cap.
The first board has also a link to be added from pin 3 to pin 5 on U2

I have the rev b board but I was looking at the rev c schematic and see there are 2 extra caps (one on pin 4 and one on pin 7 on U2) as against the rev b schematic

[6666]

In that case just hold a little bit longer as I have a small update on the way, and I'll see to make it myself. I replaced the Schmitt NANDS with a more common CD4011. I realised that I have a spare gate that calls for making a more controllable Schmitt from a pair of them. A small problem with CD4093 is that the hysteresis voltage is not directly proportional with the supply voltage, and you can't know how large it actually is for a given chip.

Ok holding on.

My NJM2068 chips arrived today will try them asap.

[johnandles]

I have the rev b board but I was looking at the rev c schematic and see there are 2 extra caps (one on pin 4 and one on pin 7 on U2) as against the rev b schematic

The sync cap needs to be added on the back of the board.(pin 7, U2)
The 'other' cap (c21) Iam not sure .(pin 4).
Hopefully George will set this right.

[6666]

Has anyone tried using a matched set of 470nf for C13 C14? I had 6 of these on hand but couldn't match two.. Amazing the spread of values 447 to 483nf from same batch.

There is a big spread amongst a hand full of caps I matched 2 @ 467nF but they are not in the board yet.

[Qiaozhi]

In the schematic of Rev C there is a 100pf cap from pin 4 on U2 going to collector of Q2. Is this another mod as it's not on pcb layout.
Also is the addition of the sync cap on pin 7 the only difference in circuit between rev c and rev b?

The sync cap needs to be added on the back of the board.(pin 7, U2)
The 'other' cap (c21) Iam not sure .(pin 4).
Hopefully George will set this right.

Oops, that's a bit of a mystery ... probably the result of some late night working.

C21 in the schematic should not be there! As noticed by CAS, it's not on the PCB, or in figures 3 or 4 of the Build Document. That's another one for the errata list.
C20 is the sync cap, and needs to be added under the PCB. This cap will be added in the next board revision.

Hope that's clear.

[Qiaozhi]

TL072 in place of the NE5532 shifts the threshold further up the adjustment pot and really smooths out the threshold. The threshold is far more stable and sensitivity is same to maybe a touch less. The chattiness as you advanced the threshold pot past the slow ticking sound is gone!

Interesting!
I'll try that mod later and see if I get the same results.

[Davor]

Guess it was "TL072 in place of the N̶E̶5̶5̶3̶2̶" should be TL062 instead. I tried several things with an integrator in sims, and my choice would be LF353 (as mentioned earlier), and also a small addition that I may suggest later on when I try it in practice. There is a looming problem of charge injection spikes passing through to the TL061 with the motion mod. There is nothing to iron them out, and TL061 is certainly not fast enough to fix them with a feedback capacitor. Ergo chatters.

[josewashere]

Let me know how you go with that one. I used a TL072

[josewashere]

let me know what you find! If good I will buy a handful to try to get a matching set

[Davor]

You are waaaay ahead of my build, but I tend to analyse things before going to actually building something (I seldom have a chance to do so), so when I do I apply mods instantly.

My Spice sims show that the differential integrator suffers from charge injection spikes originating from the FETs. Faster op amps cope better with spikes, hence TL072 and LF353 are much better choices than TL062. They both have much better noise performance. The most pronounced kind of noise here is 1/f which is indistinguishable from random drift, and it also appears as chatters.

The charge injection is unfortunately not completely symmetric and thus there will always be some offset related to it. Reducing the charge injection effects can be attempted by slowing down gate control voltage transitions. A no-brains solution is simply adding a resistor in series with a FET gate, but the value will depend upon a FET choice. With J113 I'd go with 22k to 47k, and with BF245B (my choice) which is faster and has less capacitances I'd go with ~100k. It reduces and slows down the spikes to ~500ns so that fast op amps may handle them.

The motion compensation as done in Rev-B (C15R29) is IMHO wrong. The spikes are propagated to TL061, again a slow op amp, and its feedback capacitor just can't cope. The original minipulse has a RC LP filter there, and spikes are effectively ironed out. The nature of the integrator in minipulse is so that it actively returns to zero, by way of a slew rate slope (controlled by R22), and not an exponential ever only approaching zero RC filter slope. Hence it is fast and accurate, and it doesn't require any motion filtering. I expect far less masking (better recovery speed) without the motion compensation, and also better stability if it is converted back to a LP filter with ~30Hz corner frequency.

[josewashere]

I think I am going to order another board and implement some of these changes of yours. Probably be a 3 or 4 weeks before I have it so hopefully by then you have built yours with these changes.
I do like the SAT though.

[Qiaozhi]

Oops, that's a bit of a mystery ... probably the result of some late night working.

C21 in the schematic should not be there! As noticed by CAS, it's not on the PCB, or in figures 3 or 4 of the Build Document. That's another one for the errata list.
C20 is the sync cap, and needs to be added under the PCB. This cap will be added in the next board revision.

Hope that's clear.

In the end there was no need to update the errata list, as the Build Document is not affected. It seems that I must have posted the wrong schematic, which erroneously contained C21.

The correct REV-C schematic is now available for download here -> Minipulse Plus REV-C

[Qiaozhi]

TL072 in place of the NE5532 shifts the threshold further up the adjustment pot and really smooths out the threshold. The threshold is far more stable and sensitivity is same to maybe a touch less. The chattiness as you advanced the threshold pot past the slow ticking sound is gone!

I've just tried a TL072 in place of the NE5534, and can confirm that it does smooth out the threshold somewhat, but there is a slight loss in sensitivity. Whereas as TL062 completely smooths out the threshold, and also kills the sensitivity by about 50%.

[josewashere]

Yep that is exactly what I noticed here too.
One more thing to think about.. when using the ne5532 if you place your finger on top of u10 vco 555 the threshold will decrease and sensitivity goes down noticeably even if you compensate with the threshold control! When using the tl072 and touching the 555 the threshold rises a bit instead.
When I say touching the 555 timer I mean resting the finger on the top of its case (not touching any pins) Not sure if its a capacitive effect or just coupling extra noise into the 555 but I thought it was a bit strange for a chip operating as an audio vco to behave like this and there could be a bit of gain to be had by reconfiguring this timer layout or components around it?