Replies to Minipulse Plus

[sinclairuser]

Fair enough.
I'm considering a few minor mods off the bat...
I'll replace TL062 with LF353 (or TL072) in integrator. TL062 sucks at noise and slew rate. Samples taken there are considerably shorter than the time between pulses, so the noise gets worse for that ratio. Bigger problem are the charge injection pulses due to FETs gating that may require some extra speed, as with fast logic and slow TL062 the inverting side of an integrator is at threat of producing noisy spikes.
I'm also considering BF245B that have lower gate capacitance - for the same reason - and perhaps some gate resistors too.
MC14538 could be replaced by CD4098 that has slower transitions on outputs. Faster logic produces sharper spikes, and we do not need that much speed anyway.
I'll go for a GB timing solution, and slow wired OR logic at output - that'll do.

hi davor, silverdog has included tl072 in the kit instead of 062, you might want to look at the data sheet for the tle2072, i have had great success recently substituting them for 072's in my amps(even upgraded a couple of old ones), there a modern version and totally compatable, side by side the data sheets speak volumes, they are giving them away as samples as direct from ti, the tle2071, and tle2074's are just as good too alot better specs than the old tl072 etc.

http://www.ti.com/lit/ds/symlink/tle2072.pdf

[Davor]

Fair enough.
I'm considering a few minor mods off the bat...
I'll replace TL062 with LF353 (or TL072) in integrator. TL062 sucks at noise and slew rate. Samples taken there are considerably shorter than the time between pulses, so the noise gets worse for that ratio. Bigger problem are the charge injection pulses due to FETs gating that may require some extra speed, as with fast logic and slow TL062 the inverting side of an integrator is at threat of producing noisy spikes.
I'm also considering BF245B that have lower gate capacitance - for the same reason - and perhaps some gate resistors too.
MC14538 could be replaced by CD4098 that has slower transitions on outputs. Faster logic produces sharper spikes, and we do not need that much speed anyway.
I'll go for a GB timing solution, and slow wired OR logic at output - that'll do.

[Qiaozhi]

I´d like to express my deepest sympathies in respect to the very early death of the syncronisation in MPP. R.I.P. sync! Only the good die young.
I feel somehow guilty. Next time I keep my mouth shut...

Murderer!

Seriously though ... please feel free to find problems and make improvements. Davor is already working on a GB addition, which will be interesting. The design is now open source, and I'm just acting as the project coordinator.

By the way, mistakes are things that happen to people who are doers. The only person who never made a mistake, never made anything.

[Qiaozhi]

Most probably +5V should come back if C3 is beefed up to 100u or so. Datasheet shows efficiencies down to 1kHz, and in Surf PI PSU there is an ICL7660, which is basically the same thing, and it uses 220u.

The ICL7660 drops out at a much higher voltage (I think it's around 10.5V) which is why I chose the LT1054. It also seems to generate much less switching noise. If (in the future) we upgrade the MPP to include a micro-controller, it will be possible to synchronize the voltage inverter oscillator at a higher frequency which is harmonically related to the TX pulse rate.

Please feel free to experiment with the current setup and see if you can get the sync working. But but sure to check it over the full range of power supply voltages.

Remember ... the MPP design is my [late Christmas] gift to the Geotech Forum, and you are all welcome to modify the design to meet your own requirements. If/when the design gets updated to include new features, then some of these modification will most likely be added. Like Linus Torvalds with the Linux kernel, I reserve the right to be the final arbitrator on what mods get included.

[Davor]

Larger capacitor pours more juice. In case of higher frequency and smaller capacitor, it pours less juice more often, so I think it should do. Schottky diodes are a better choice for smaller drop, and in this case their leakage does not matter.
Synchronisation is beneficial for lower noise, so I'd go with sync.

[mickstv]

I use a LT1054 as a charge pump in the detector I built and the cap I use is a 220uf (in the Minipulse it's C3). I also use Schottky diodes in the voltage doubler (in the Minipulse it's diodes D1 and D2). Schottky have a lower voltage drop.

[sinclairuser]

that funny davor, when i was looking at the schematic yesterday i wondered how long it would be before someone tried a 7660s.
do you think the bigger cap will re-impliment the synchronisation?, i liked the idea of that.

[Davor]

Anyway, after a quick repair, I added the missing connection from Q2/R5 to U2 pin 7 ... and it kills the +5V supply.

Most probably +5V should come back if C3 is beefed up to 100u or so. Datasheet shows efficiencies down to 1kHz, and in Surf PI PSU there is an ICL7660, which is basically the same thing, and it uses 220u.

[migusch]

[...] I added the missing connection [...] ... and it kills the +5V supply. [...] which unfortunately means we cannot sync the TX osc to the voltage inverter. [...]

I´d like to express my deepest sympathies in respect to the very early death of the syncronisation in MPP. R.I.P. sync! Only the good die young.
I feel somehow guilty. Next time I keep my mouth shut...

[sinclairuser]

Shield connection goes to 0V, and you should shield both Tx and Rx.
Foil is not the perfect material for the job though, and if you use it remember to insulate Tx shield from Rx shield, and connect them only in one point to avoid loops.

thanks davor, thats what i thought about the coil, the foil is in reference to the foil shield in the 1266x case, since i'm using this case i thought it may help with emi, i'll be using graphite as usual.
@ george, thanks, these things are sent to test us.

[Qiaozhi]

Having a closer look at the schematic (minipulse+ rev-B) I wonder why synchronisation of LT1054 to the oscillator 555 is made via pin 6 of LT1054 ("Vref") and not via pin 7 ("osc")?

Nice catch
There is a missing connection between a collector of Q2 and a pin7, and it is easily corrected with a piece of wire, like this:

I've been looking at the sync problem this evening, and ...

Firstly, the battery pack I was using to test the MPP was getting low, so I grabbed a fully charged 12V lead-acid battery (7Ah) and connected it via jumper leads. Big mistake! Accidentally got the battery leads reversed and [literally] let the magic smoke out of the -5V regulator (U1). (Doh!) The only other component to R.I.P. was the TX 555 oscillator (U4). At least you now know which components are going to kick the bucket if you reverse the battery leads.

Anyway, after a quick repair, I added the missing connection from Q2/R5 to U2 pin 7 ... and it kills the +5V supply. No doubt this is due to the much lower frequency of the TX oscillator compared to the 20kHz internal oscillator of the LT1054, which unfortunately means we cannot sync the TX osc to the voltage inverter. However, the diode pump circuit of the original Minipulse was also free-running, and it worked just fine. In fact I cannot see any disturbance on the pre-amp output due to switching noise.

The conclusion of my investigations is that there is no requirement to fit either R5, R6 or Q2.
I will remove these in a future update.

[Davor]

Shield connection goes to 0V, and you should shield both Tx and Rx.
Foil is not the perfect material for the job though, and if you use it remember to insulate Tx shield from Rx shield, and connect them only in one point to avoid loops.

[sinclairuser]

hi all, i have been designing my coil for this and decided to build a split tx/rx ib style, and its got me thinking, do i need a shield on the tx?, or should i shield both coils?.
also does the +ve shield connection have any considerations?, more over i'm using an old 1266x case, if i employ the foil shield where does it connect?, i suspect i know the answer, but i dont want any mistakes on this.
thanks,
ally.

[Qiaozhi]

Having a closer look at the schematic (minipulse+ rev-B) I wonder why synchronisation of LT1054 to the oscillator 555 is made via pin 6 of LT1054 ("Vref") and not via pin 7 ("osc")?

Oops! ... This looks like a mistake. I'll check it tonight and confirm.

[Davor]

Having a closer look at the schematic (minipulse+ rev-B) I wonder why synchronisation of LT1054 to the oscillator 555 is made via pin 6 of LT1054 ("Vref") and not via pin 7 ("osc")?

Nice catch
There is a missing connection between a collector of Q2 and a pin7, and it is easily corrected with a piece of wire, like this: