HH2 -- first cut

[Carl-NC]

There is an obvious error in my schematic, R7 should not be there. Also, for fastest speed R8 & R15 should be zero.

But bipolar will still work just fine, as long as you pay attention to details. With R8 & R15 = 0, there are still 2 speed limitations in my driver circuit... what are they? Hint: Aziz's low-cost variant addresses one but not the other.

- Carl

[Aziz]

Hi Carl,

it seems, that nobody besides me is interested on your speed limitations. At the moment, I am currently again on this topic and have some more improvements which addresses also the battery power efficiency.

I will post this as soon as I am finished with analysis.

Regards,
Aziz

[Aziz]

Hi friends,

the improvements addresses the following features:
- softer (slower) TX switch-on producing lower ripple noises in the electronics during switch-on, while keeping the fast switch-off feature

- faster switch-off achieving quicker the breakdown voltage of the MOSFET (fast dI/dt)
We can saturate (energize) the target with slightly smaller transmit pulse width. This will save battery power a little bit.

- driving higher Cin MOSFETs (input capacitance) or parallel stacked MOSFETs (for higher coil current flow or lowering the total Rds-on).

The common available integrated MOSFET drivers can not achieve the features above as they are optimized for both switch-on and off. The design proposal is a slight modification of the low-cost variant shown already above. The switching latency delay isn't important here. I will post the schematics soon until I am finished with the simulations.

Aziz

[barry]

Look forward to your findings aziz. Thanks for sharing.

[porkluvr]

Hi Carl,

it seems, that nobody besides me is interested on your speed limitations. Aziz

I'm interested, but I'm relatively clueless. In trying to adapt the Aziz drive to a LM555 output, it occurs to me that the Deltapulse and the (original)Hammerhead have opposite drive logic. Why didn't I notice that before... duh.

Well, if this was easy then everybody would be doing it.

Let me spend a few more days bogged down in this small detail. Then maybe I'll get an inkling.
***************************
I've started with the DP transmitter and managed to get a complex waveform (1 long + 3, 4,or 5 short pulses). Now - if I can work this in and maybe come up with a blanking gate, then I'll eventually do some prospecting.

[Aziz]

Hello friends,

here is the lowcost PI MOSFET driver spice model circuit for all, who wish to experiment on this and find the issues rised from Carl. This model does not contain the improvements yet.
I have rebuilt the lost model again and put this here for download (good for backupping). Use the latest Linear Technology LTSpice IV software in "Alternate" mode (Tools/Control Panel/SPICE tab/Engine Solver: Alternate).

Happy finding...

Aziz

[Stefan]

Aziz,

If you are using a TCS426/7/8 MOSFET drivers there is a simple solution to give slow turn on and fast turn off.

Slow turn can be adjusted by placing a 10 to 20 ohm resistance in series with the gate drive. This will add a small time constant to the charging of the gate for on time. ie slow turn on.

By placing a diode in parallel with the resistor and biased such than on turn on the current is flowing through the resistor. At turn off the diode is forward biased and will discharge the gate charge quickly. The resistor has no effect at this point. ie fast turn off

I have used this method and it works fine.

Regards,

Stefan

[Aziz]

Hello Stefan,

you have found the soft switch-on solution.

A resistor must be placed in the gate path. The reverse biased diode parallel to this gate resistor gives not much an improvement on the modified low-cost variant without the BJT. Even then, when the drain resistor of BS250 is kept low.

With the MOSFET driver IC's, the switch-off could be fast due to low impedance discharge path. But I want to omit any driver IC's (should be a low-cost design).

Aziz

[Aziz]

Hello friends,

here is the slight improvement of the low-cost PI MOSFET driver. By stacking the 2N2907 (Q1 paralleled with Q2), the circuit is able to switch off faster, be able to drive more paralleled power MOSFET's or a higher input capacitance MOSFET's. If you are using only one power MOSFET, then only one 2N2907 is enough. The resistor R3 gives a soft-on feature. I have reduced the R2 to make the Q1 faster for switch-off. The higher switching current through BS250 is a peanut compared to the potential for reducing the switch-on time due to faster switch-off of the power MOSFET. The switch-on time is delayed due to R3 to a few microseconds. The switch-off delay is just around about 200 ns. The coil current will be switched off in about 70ns until it reaches the avalanced state of the MOSFET.

If you stack the power MOSFET's, then ensure both devices are mounted on the same heat plate to have same temperature. Other PNP transistors for Q1 could be used. Just try it out.

I will enclose the whole spice simulation model for further experiments for you. Look for the improved version in the zip-file.

Why I did this effort?
I couldn't get the MOSFET driver IC's on my local shop or supplier! But all the parts used in the low-cost variant are most common, available and cheap.

Happy simulating..

Aziz

[porkluvr]

Aziz, I have a feeling that the BYV28-200 you are using would not stand up very well in a circuit using the IRF740. I had a hard time finding models for 400V diodes, but found models (and suppliers) for 600V (fast recovery) diodes from Vishay and Fairchild.

Some of the models should be added to (LTspice) standard.dio and the subcircuits are to be associated with your x-diode symbol.

These should work with any PSpice compatible simulator.
When I get some more time maybe I'll look for some 400V diodes but it is not a priority.
********************
I might abandon the series diode because I sometimes see a 400uV glitch in my listening period. I haven't determined exactly what circuit combination is causing it but it's something I can't ignore. Rather than pull my hair out trying to analyze it I may just pfft and use a resistor.

What is that series diode there FOR, anyway, do you know? What is the advantage of having a fast recovery diode between the FET and coil?

[Aziz]

Hi porkluvr,

interesting incident. Your supplied diode models are coming from my living town (Rosenheim). The BYV28-200 is the one I found and did not search for better versions yet. Generally, the higher the breakdown voltage, the higher the drop voltage and thus reducing the power efficiency of the transmit pulse. A bad replacement for the BYV28-200 could be a 1N4004 - 1N4007. So it might be a good task for all, to find better replacements (+ spice models). I don't know, whether the BYV28-200 will withstand 400 V. Any input will be appreciated. The BYV28-400 is also available but I haven't any spice models for it.

The main function of the diode is to reduce the MOSFET's capacitance affecting the transmit coils capacitance. After the avalance breakdown voltage is falling short of, the diode is becoming a capacitor which is connected in series to the MOSFET's capacitance. The total effective capacitance will therefore be lower than the diodes capacitance and has less capacitive load to the transmit coil. The transmit coil can be damped then faster which results in a higher damping resistor.

When I have more time, I will look for better diodes with spice models. But I do not edit or modify the standard files in LTSpice (due to update). So it is better to make external spice models. Thanks for your spice models and hint.

Aziz

[okantex]

Hi Aziz ,

how are you ?

I am sure you remember that I am studying to make TX part for QED.BW's homework to me.
I want to use your improved driver ,but I could not understand how to control it.
in schemetic what is V2 ?
I can understand Carl's and BW 's drawings , but can not understand your simulation drawings.
according to your improved schematic ,how can I control TX.
where Should I put TX timing pin ( 18f4550 's control pin for TX (driver pin))?
Btw I want to use 18F4550 for QED. how can I do connection.
can you explain a little more ,please
regards
Okantex

[Aziz]

Hi okantex,

Hi Aziz ,

how are you ?

I am sure you remember that I am studying to make TX part for QED.BW's homework to me.
I want to use your improved driver ,but I could not understand how to control it.
in schemetic what is V2 ?
I can understand Carl's and BW 's drawings , but can not understand your simulation drawings.
according to your improved schematic ,how can I control TX.
where Should I put TX timing pin ( 18f4550 's control pin for TX (driver pin))?
Btw I want to use 18F4550 for QED. how can I do connection.
can you explain a little more ,please
regards
Okantex

The mosfet driver circuit is tailored for the negative supply voltage operated micro controller and coil. The coil flyback voltage will give a positive voltage response. If your micro controller and your coil supply has not the same supply configuration, then you can not use the schematics. QED uses a different supply configuration I think. So the schematics must be adapted.

V2 is the transmit trigger pulse from the micro controller and is used only in the simulation model for clearness. The micro is operated from system GND = micro VDD and system -5V = micro GND. The mosfet BS250 is for logic level shifting from 0/-5 into 0/-Ucoil (0=ground). It is also inverting the pulse logic (simple inverter). -Ucoil can be much more (e.g. -12V). A zero volt on BS250's gate means transmit coil off and a -5V means transmit coil on. So the micro has to support a negative transmit logic: I/O pin=0 => transmit pulse on, I/O pin=1 => transmit pulse off.
Giving a higher switching voltage to the IRF740 will lower the mosfets on-resistance. Therefore the full coil voltage rail is fed into IRF740's gate. Care must be taken to the two 2N2907. They should be matched together (= same characteristics). If not, they need an additional seperate resistor on each base. Usually one 2N2907 is enough to discharge the IRF740's gate charge quickly.

If the micro has tri-state I/O output on the transmit pulse pin, an additional resistor (e.g. 100k) from the gate of BS250 to source pin must be connected to force switch-off of the coil. Otherwise, it could have an undefined state and could switch the coil on/off arbitrary.

I am not familiar with PIC's. I don't like mature PIC micros. I hate their banking (paging) and inherent inefficiency of coding. The only PIC i could like would be a dsPIC33F..

Aziz

[okantex]

Hi Aziz ,
I can not program atmels ,I do not know assembler or C++.so I have to use pic.

I could not understand your below words

"
zero volt on BS250's gate means transmit coil off and a -5V means transmit coil on. So the micro has to support a negative transmit logic: I/O pin=0 => transmit pulse on, I/O pin=1 => transmit pulse off.
"
I/O = 0 => transmit off or on.?

thanks


so if my pic can give -5 when it is on ,I can drive your tx circuit.

[barry]

Hi Aziz ,
I can not program atmels ,I do not know assembler or C++.so I have to use pic.

I could not understand your below words

"
zero volt on BS250's gate means transmit coil off and a -5V means transmit coil on. So the micro has to support a negative transmit logic: I/O pin=0 => transmit pulse on, I/O pin=1 => transmit pulse off.
"
I/O = 0 => transmit off or on.?

thanks


so if my pic can give -5 when it is on ,I can drive your tx circuit.

micro output register set to 0
voltage at Micro output pin is -5volts
BS250 turns on
Coil Transmit on

micro output register set to 1
voltage at Micro output pin is 0volts
BS250 turns off
Coil Transmit off