Bipolar PI design

[dbanner]

You can, see the Pulse-1 design in ITMD3. However, the 4066 has advantages:

• In a multi-channel design the switches will match reasonably well which is a bonus.
• The JFET characteristics are more dependent on preamp signal levels.
• The JFET requires clock translators in the Bipolar-PI design because the micro runs on +5V/0V.
• Without translators, 4 discrete transistors take up about the same space as a 4066 chip. With any more components, the 4066 wins.

"clock translators" the correct term? You mean a -5V swing is needed which the micro cannot give.because it runs on +5V/0V
4066 definitely wins with so many channels. No doubt about that
Also easier to route on pcb . 4 channels of transistor switchers and other components I imagine can be silly waste of space.

[Detectorist#1]

Before using of J113 transistors, I measure Rds with ohmmeter. Usually Rds is in range 40-80 ohms. This measuring have to be made with short connection between Gate and Source. For using, I take a couple with no more than 10 ohms difference. This is 1% of the value of serially connected resistors in scheme whose popular value is 1Kohm.

[eduardo1979]

Thanks Carl for your comments. I will have to make the modifications to the scheme, but it has to be for this weekend since I don't have enough time for now.

[Carl-NC]

Any reason why FETs can't be used instead of 4066? Can they not be switched directly by the pic. Seems like much less components.

You can, see the Pulse-1 design in ITMD3. However, the 4066 has advantages:

• In a multi-channel design the switches will match reasonably well which is a bonus.
• The JFET characteristics are more dependent on preamp signal levels.
• The JFET requires clock translators in the Bipolar-PI design because the micro runs on +5V/0V.
• Without translators, 4 discrete transistors take up about the same space as a 4066 chip. With any more components, the 4066 wins.

[eduardo1979]

Surely it will be in SMD. For now we are going to wait for detectorist#1 to finish its work, so we can include it in the final schematic. There is no rush, good jobs come when they are done slowly. Questions always arise and I hope that people are encouraged and give their opinion.

[Carl-NC]

Eduardo, here's my feedback:

1. I generally agree with Detectorist#1 on the preamp, however you often have to test these things to see whether 1 stage or 2 stage is faster. You can't go by GBW alone as overvoltage recovery plays a strong role, and that can get worse with high gain. I built a little test circuit that just has a monopolar TX and the preamp that I use for testing different opamps with different gains with both 1 stage & 2 stages.
2. The LMH6646 is serious overkill for U8b. Since this is a unity-gain amp anything with several MHz GBW will work.
3. You've combined the demod input resistors into R38 & R39. Problem is, if (for example) U9d and U9c accidentally turn on at the same time then U7b and U8b outputs are shorted and they burn up. Trust me, this can happen, especially when debugging buggy software. I like to place the resistors on the preamp side of the switches to avoid this. Also, it places the 4066 switches directly on virtual ground so they behave better versus preamp signal swing.
4. I didn't do this in the Pulse5 design, but you may want to add an offset resistor between P3 and U11b. As it is, the GB pot adjusts from purely TGT only to purely GND only. You should not need this much GB range so the offset resistor can be added to reduce the range and make it more reasonable. You'll have to figure this out later as it depends on timing.
5. D13-D14 don't really do anything but add a dead zone at zero volts. This can be useful for reducing noise but it also reduces sensitivity. You can leave them in the design but try shorting them when testing.

All for now.

[algan]

Hello Eduardo, what variant is it, do you draw the circuit board dip or smd?

[Detectorist#1]

Hi Eduardo,
Now I'm in intensive testing of the schematic and settings of additional module for almost all simple PI designs for sure Fe/non-Fe discrimination. The results will be posted here.

[eduardo1979]

Hello Detectorist#1.
To power the detector I would use a 16.8V/3400mAh LI-ION battery pack. I have used this type of battery for many years and they give very good results. I would leave the 10V voltage regulator there, because it helps a lot when the voltage level of the batteries drops significantly. In order not to lose depth, I would not increase the voltage, but rather try to increase the pulse width and leave the voltage at 10V. LT1112 operational amplifiers in SOIC8 format I have more than 50 units bought at Mouser a year ago and I would like to use them in my projects, since they cost me a lot of money. That is why I used the LT1112. I think your idea of using a single stage in the preamplifier is good and thus we save some additional components. I attach the corrected schematic. I really like your idea of adding an additional channel to be able to discriminate between ferrous and non-ferrous. I would like you to make a drawing so I can add it to the original scheme.

[Detectorist#1]

Hi Eduardo,
Also dividing of the first RX stage in two stages with OPA1612AID not improves useful MIN delay in PI design. This idea is useful in old designs with NE5534 - NE5532. OPA1611AID is fast enough for real fast RX stage at real practice with additional sources of problems as ground and wet grace influence.

[Detectorist#1]

Hi Eduardo,
Your variant of the schematic needs to use in decoder's part OpAmps with very little Vos. This is true for LT1112 but convenient case DIP8 is obsolete. And SOIC8 case for DIY projects is not convenient and the price in MOUSER is 10.31USD 1+. Every real project is full of compromises!
BTW, for improving of the stability of the decoders in original schematic of Carl, 10K resistors have to be added in IC7c and IC7d in series of diodes D9 and D10 to the negative inputs of OpAmps.

[Detectorist#1]

Hi dbanner,

Yes, have at least one reason against using of J-FET as J113 as switches in similar projects - the voltage Ugs must not to be more than +/- 0.4V when switch have to be in ON state. Plus 0.4V is the border of conducting of a current on direct direction of P-N junction. Minus 0.4V is the border when P-N junction (gate-source) of J113 transistor starts to be closed - but the switch have to be closed in this situation. In short - it is possible to use J113 as good switch only if Ugs in ON state is not more than +/- 0.4V. In some projects this is possible but not in all projects. Using of 4066 not suffers from this problem.
Four transistors haves total 4x3=12points in PCB. CD4066 haves 14points in PCB. Not serious reduction for PCB tracing.

[dbanner]

Any reason why FETs can't be used instead of 4066? Can they not be switched directly by the pic. Seems like much less components.

[Detectorist#1]

Hi Eduardo,
Yes. I agree with you - this is great project from Carl. Also I congratulate you for efforts to improve this project. Carl have to be satisfied - the aim of ITMD3 is in action - we together made efforts to design more good DIY projects!
My comments for your variant of the schematic:
1. First, 16V power supply source is very inconvenient. How you imagine this source? 10 pcs of AA batteries starts as 15V but have to be used until 1.0V per cell - 10x1=10V and the 7810 stabilizer will not work totally. The variant with 4xLiPo accus starts with 14.4V and ends as 4x3=12V. But this variant needs special charge source with voltage equalization - expensive option. Now, I design some variants for PI MD supply with using of two LiPo cells and DC-DC converters from 6-7.4V to stable 12V. At this moment my design have 84% efficiency. I continue to look for the solution with improved efficiency.
2. The original schematic of the demodulators is a bit problematic as you mention in post #63 and my post #67.
3. For adding of 10V stabilizer - at first look this stabilize supply voltage of the TX part -good idea. But the deepness of target's recognition depends in PI with power of 6 from TX power. In reality - reduction in TX power voltage as the voltage of the batteries go down, influences very little on the deepness (next compromise in the design of PI MD).
My dream at last time - to add to this design additional channel with TX ON-time analysis for sure Fe/non-Fe recognition. Have to be used balanced additional RX coil.

[algan]

Hello, colleagues, what programmer do you program the chip with, I tried a little way with pickit3 Chinese version, everything was successful, then I tried to launch only the controller, but there are no pulses, which programmer would you recommend. .Thank you.