MONOCOIL SUCCESS

[moodz]

Hey Moodz, it is called DISCRIMINATION, not Descrimination :P

Sorry I are an engineer.

[nick_f]

Descrimination is available However it is via a two sample method which is not quite as "neat" as the single sample used on the differential coil.

However it is a vast improvement over conventional monocoil front ends.

moodz

Hey Moodz, it is called DISCRIMINATION, not Descrimination :P

[moodz]

Moodz with your blocker invention
you say we dont need a high voltage fet switch
I have some logic level fets good for 30v and 55v
do you think they would work ?
thanks

You can use a low voltage fet for the switch but your blockers must still be high voltage and able to carry the tx pulse current.
moodz

[6666]

.... here is another ( to be ) tested improvement ..... the differential head, damping and blocker circuit can be located in or very close to the search coil ... This will eliminate the flyback pulse having to travel up the coil cable to the control box ..

The blocker transistor Q2 should have a low "ON" impedance.

and again it is the most simple change ...

PS the TX switch device does not need a high voltage rating now ... the blocker handles the flyback volts

moodz

[ATTACH]17702[/ATTACH]

Moodz with your blocker invention
you say we dont need a high voltage fet switch
I have some logic level fets good for 30v and 55v
do you think they would work ?
thanks

[Midas]

I finally got the circuit built and it seems to function as mentioned except for one small problem, the frontend is a little deaf compared to a standard mono front end.


I end up taking the blocking fet circuit out and used a matched pair of 1k resistors and crobar diodes on the TX fet side then into a diff amp. My modified setup works fine, the TX is about 100us running about 400volts flyback and it's decay at the diff amp is about 6us.



Mick

Hi Mick, hope your Christmas went well. Thanks for the result. Sounds like it might be a no go... I wonder if the same is also true of the differential passive fet blocking scheme.

[mickstv]

I finally got the circuit built and it seems to function as mentioned except for one small problem, the frontend is a little deaf compared to a standard mono front end.


I end up taking the blocking fet circuit out and used a matched pair of 1k resistors and crobar diodes on the TX fet side then into a diff amp. My modified setup works fine, the TX is about 100us running about 400volts flyback and it's decay at the diff amp is about 6us.



Mick

[Aziz]

Hi guys,

no doubt, you can use the discrete version too. You just need a compensation network for best CMRR (crucial point in differential amplifier applications). This means adjusting the source impedance (source resistance and capacitance) and the amplifiers resistor network.

The complete instrumentation-amp IC's reduces noise, pcb space and cost of course.

Aziz

[Midas]

Considering that the unbalanced input of the amplifier in most PIs have very little, if any common mode rejection ( thats why you need all the elaborate shielding on the coils ) the discrete differential amplifier will be fine. You actually only need about 16 db of CMRR whereas an "unmatched" discrete differential amplifier will achieve usually better than 30 db.

The bottom line is that for this application you dont need some fancy pants CMRR rejection level .. you will never be satisfied with the performance if you indulge in catalogue engineering ( there is always another chip with better specs ).

Good engineers make what they want out of what they can get.

Make it work then make it better .. not the other way around.

moodz.

Hi Moodz,

I'm not sure what you mean when you say you don't 'need' fancy CMRR. Are you saying that past 16db you stop seeing any benefit in noise reduction?

I was thinking the more CMRR the better which is one of the reasons I decided to go with the discrete opamp solution. My plan was to add in a multi-turn trimpot to one or two of the resistors where I was hoping I'd be able to to adjust it its to a better CMRR than the THAT1510. Perhaps its not worth it.

So what did you mean when you said:
"The THAT amplifier is not the best here ... use opamp DC diff amp."
?

Midas

[moodz]

Considering that the unbalanced input of the amplifier in most PIs have very little, if any common mode rejection ( thats why you need all the elaborate shielding on the coils ) the discrete differential amplifier will be fine. You actually only need about 16 db of CMRR whereas an "unmatched" discrete differential amplifier will achieve usually better than 30 db.

The bottom line is that for this application you dont need some fancy pants CMRR rejection level .. you will never be satisfied with the performance if you indulge in catalogue engineering ( there is always another chip with better specs ).

Good engineers make what they want out of what they can get.

Make it work then make it better .. not the other way around.

moodz.

[Aziz]

Hi all,

I personally wouldn't use the discrete instrumentation amplifier version. Any imbalance of the source impedance (source resistance, source capacitive load) and/or the used resistors in the discrete version (requires better than 1% tolerance), will radically reduce the common mode rejection ratio (CMRR). Particularly much more, if you have high frequency signals (no doubt, we have it here). The CMRR is frequency dependent!!!

The money is well invested in an instrumentation amplifier IC (laser trimmed internal resistors).

Aziz

[moodz]

so you reckon that would be the way to go Moodz ?

Well what I meant was you could use them if you have them on hand .... they are fast enough and reasonbly low noise / low cost ... depends if you demod then amplify or amplify then demod.

moodz

[6666]

I reckon just make up a diff amp using ne5534 ... ???

so you reckon that would be the way to go Moodz ?

[mickstv]

The input diodes are not a show stopper .. for instance if you run the front end off 8 volts the voltages on the output of the blocker will be from 1 to 17 volts approximately ( due to leakage of the high voltage flyback ) ... so you would want to run the THAT off say 0 and +20 volts to give you a bit of headroom. The 8 volt supply is the analogue ground reference point. The THAT can be powered off upto +-20 volts ( ie 40 volts ) so if you did have a 40 volt supply to power the THAT you could run the frontend off something as high as 18 or 19 volts ... As a rule of thumb the zener diode on the blocker should be twice the supply voltage but not higher than the Gate breakdown voltage of the blocker mosfet.

with an 8 volt supply and the THAT running from +20 volts the protective diodes were never forward biased.

PS you must make sure the damping is correct so the coil flyback does not cycle negative either ....

moodz

Thanks for the info.



Mick

[Midas]

The LME49990 certainly has some pretty good looking specs but it is an OP-AMP not a diff AMP like the THAT .... snippet from the 49990 spec sheet below if you want a diff amp ... I am not saying dont use it though ...( if you got em ... smoke em )

Yep I realize that. That was pretty much the circuit I was planning on using. I thought we'd already given up on single chip diff amps. But if you do know of\find one with specs as good as the lme49990 I'm all ears. I'm sure I can find another use for the lme's. Whatever I use though I'm actually hoping on keeping the magic smoke inside the little black box.

[moodz]

The input diodes are not a show stopper .. for instance if you run the front end off 8 volts the voltages on the output of the blocker will be from 1 to 17 volts approximately ( due to leakage of the high voltage flyback ) ... so you would want to run the THAT off say 0 and +20 volts to give you a bit of headroom. The 8 volt supply is the analogue ground reference point. The THAT can be powered off upto +-20 volts ( ie 40 volts ) so if you did have a 40 volt supply to power the THAT you could run the frontend off something as high as 18 or 19 volts ... As a rule of thumb the zener diode on the blocker should be twice the supply voltage but not higher than the Gate breakdown voltage of the blocker mosfet.

with an 8 volt supply and the THAT running from +20 volts the protective diodes were never forward biased.

PS you must make sure the damping is correct so the coil flyback does not cycle negative either ....

moodz