Replies to Minipulse Plus

[Davor]

As mentioned before, I'm considering replacing TL062 with LF353, and adding some 100k in series with the FETs gates. I made a simulation in LTspice that clearly points to the sources of charge injection. The attached project contains lots of different opamps and many more FETs. I also checked TLE2072 model, and it shows considerable improvement, but I can't buy it here. Besides, LF353 has superb 1/f noise.

[sinclairuser]

As mentioned before, I'm considering replacing TL062 with LF353, and adding some 100k in series with the FETs gates. I made a simulation in LTspice that clearly points to the sources of charge injection. The attached project contains lots of different opamps and many more FETs. I also checked TLE2072 model, and it shows considerable improvement, but I can't buy it here. Besides, LF353 has superb 1/f noise.

thanks for the file davor, its a shame about the tle2072, i will try it my end if it works out i could allways post you a couple of samples.
if we try different approaches, we will know what works and what doesn't.
i for got to say before to everyone dont use a tle2072, unless you have a tle2071 to replace the tl061, when your going for lower noise choices like here, replacing one chip is a waste since noise could come from the 061, i have found these newer versions benefit from this approach.

[Davor]

Chip replacement with a more contemporary ones is often working good, provided a new chip is in fact better for the job. For this function we need a FET input op amp with noise performance better than 100nV/sqrt(Hz) at 1 Hz, and over 20V/us slew rate. As low the 1Hz noise, and as high slew rate, the better. TLE2072 is in that sense somewhat better than TL072, and less so against LF353, as LF353 has less noise.
I found a distributor that will send me a TLE2072 (eventually) that I'll compare with other chips, but for the time being, LF353 will stay.

I'll concentrate my efforts to building the device and a coil, and to designing a GB board ... unless someone beats me to it. I've let all the details necessary for making it.

[silverdog]

more REVB boards have arrived
I have updated the shop stock levels for pcbs and kits
As this was a bigger batch of pcbs and the GBP is strong against the USD at the moment they have worked out cheaper

[Qiaozhi]

more REVB boards have arrived
I have updated the shop stock levels for pcbs and kits
As this was a bigger batch of pcbs and the GBP is strong against the USD at the moment they have worked out cheaper

Please note that the new set of boards are actually REV-C.

This revision has fixes for the two small errors mentioned in the Build Document (Step 2 and Step 9), and an additional testpoint where you attach a ground probe. However, the recently discovered sync issue is not fixed, and requires you to not fit R5, R9 and Q2. These components will not be included in the Silverdog kits.

I will update the Build Document shortly and add a new REV-C thread with the latest details.

[6666]

Kit has arrived, Thanks Andy.

[6666]

So far both +5 & -5 volts works ok, I must admit I am not too keen on the tiny solder pads.

If you have the kit no plugs or pins are provided for PL 1-5, you need a trip to your parts store
here is a tip, if you want plugs with wires have a look in your R/C model shop for servo leads.

[6666]

The beast lives, nice pulse at TP1.

[Davor]

That was fast

[Davor]

A few thoughts ... For a revision D (or whatever comes next) I'd like to see PCB placers for:
- a 10k resistor between U8 pin5 and GND - this will allow for using bipolar op amps at this place as well, and does not hurt performance in case of a FET op amp
- 100k resistors in series with Q3 and Q4 gates - I'll fit those "Ikebana style" on a Rev B PCB
- a 47pF capacitor can be added in parallel with R22 to stop propagation of switching glitches back to the feedback network.

I think with a good bipolar op amp that does not produce too much noise, and has a decent common mode range, the input gain could be reduced, the integrator stage gain could be increased to compensate, and the sampling could commence a bit sooner. Choice of well designed bipolar op amps is much wider than in case of FET op amps. There is no problem omitting any of these additions or replacing them with jumpers.

Replacing FETs with a bilateral switch seem a good idea. Eric recommended them as well. Small trouble is that in case of Minipulse it would require ±5V supply in a switching logic to power these properly. I can fix this on a Gb board

[silverdog]

my mistake, I will be providing pins and plugs for PL1-5

So far both +5 & -5 volts works ok, I must admit I am not too keen on the tiny solder pads.

If you have the kit no plugs or pins are provided for PL 1-5, you need a trip to your parts store
here is a tip, if you want plugs with wires have a look in your R/C model shop for servo leads.

[Qiaozhi]

A few thoughts ... For a revision D (or whatever comes next) I'd like to see PCB placers for:
- a 10k resistor between U8 pin5 and GND - this will allow for using bipolar op amps at this place as well, and does not hurt performance in case of a FET op amp
- 100k resistors in series with Q3 and Q4 gates - I'll fit those "Ikebana style" on a Rev B PCB
- a 47pF capacitor can be added in parallel with R22 to stop propagation of switching glitches back to the feedback network.

I think with a good bipolar op amp that does not produce too much noise, and has a decent common mode range, the input gain could be reduced, the integrator stage gain could be increased to compensate, and the sampling could commence a bit sooner. Choice of well designed bipolar op amps is much wider than in case of FET op amps. There is no problem omitting any of these additions or replacing them with jumpers.

Replacing FETs with a bilateral switch seem a good idea. Eric recommended them as well. Small trouble is that in case of Minipulse it would require ±5V supply in a switching logic to power these properly. I can fix this on a Gb board

The MPP will probably not change drastically from what it is now, and any future revisions will only incorporate small evolutionary fixes to correct potential errors or make small improvements.

Additions such as GB, and bipolar opamps with various optional components, etc., can be incorporated in a more advanced version (Minipulse Super Plus?).

I did read somewhere where Eric suggested that bilateral switches were a better alternative to FETs. However, the J113s have the advantage of taking up minimal PCB real estate, and make routing easier. Bilateral switches probably make more sense in a GB-enabled MPP where you can make use of all 4 switches in the package. Also, I notice the CS6 PI (which is one of Eric's more modern designs) continues to use FETs.

[johnandles]

That was fast

Ozzies in the lead so far......"Ozzy Ozzy Ozzy.....!"
Lol

[ivconic]

The MPP will probably not change drastically from what it is now, and any future revisions will only incorporate small evolutionary fixes to correct potential errors or make small improvements.

Additions such as GB, and bipolar opamps with various optional components, etc., can be incorporated in a more advanced version (Minipulse Super Plus?).

I did read somewhere where Eric suggested that bilateral switches were a better alternative to FETs. However, the J113s have the advantage of taking up minimal PCB real estate, and make routing easier. Bilateral switches probably make more sense in a GB-enabled MPP where you can make use of all 4 switches in the package. Also, I notice the CS6 PI (which is one of Eric's more modern designs) continues to use FETs.

CScope simply loves FETs, especially J113!
Hard to come by... expensive too!
MPP... any video so far?

[Davor]

J113 is a classic.
"Can't beat the classic" - Jason Statham - The Expendables 2

I prefer BF245B though. Small, cheap, fast, symmetric.
But Qiaozhi is right, I can have a complete switching solution for Minipulse GB with a single 4066, and no diode OR-s.