IRF9640 GOOD SNUBBER SCHEMATIC NEEDED

[daverave]

ok. i grasp with 3.2kHz you want to make C-Scope 6PI... but then a coil must be with 1mH value... NOT 300uH.
irfd110 has also less low R in open state... compare datasheets on both 9640 and it. so heat energy on irfd110 is very low, a coil
also has high ohm vs 300uH coil, also in serial with coil sets a resistor... all those gives low heat on the FET.
and all that was keeping in consideration by you.

i was thinking of trying a 500v mosfet but the on resistance is 6 ohm and my coil is already around 2.6 ohm.

[württemberger]

I had a similar problem with my Mirage Pi with IRF 740 (400 Volt)...
The Mirage: Freq. 1 KHz, Pulswith 50µS.
The Coil: 304µH and 1,6 Ohm.
Flyback-Puls more than 450 Volt !!!

First, i lost much searchdeep and 20 Minutes later, the IRF 740 was death...

Now i used the IRF 840 with a higher Breakvoltage (500 Volt) and
all is O.K.

The IRF 9640 Breakvoltage is only - 200 Volt.

[Qiaozhi]

Dave's problem is that (with his current configuration) he suffers a loss of depth after about 20 minutes. The problem remains even when the circuit is allowed to cool down. The only way to restore the original depth is to replace the MOSFET. So it appears that the MOSFET is getting damaged in some way. As I understand it, the MOSFET is not getting hot.

However (according to the datasheet) the IRF9640 is repetitive avalanche rated, and the maximum gate voltage is 20V. The single pulse avalanche energy can be as high as 700mJ. For example: VDD = -50V, starting Tj = 25 degrees C, L = 8.6mH, Rg = 25 ohms, Ias = -11V.

The one thing I could suggest is to add a gate resistor (e.g. 100 ohms) between the 555 and the IRF9640, as the Surf-PI has a direct connection. I'm not sure why the lack of this resistor could cause the MOSFET to fail, but it's worth a try.

[Qiaozhi]

I had a similar problem with my Mirage Pi with IRF 740 (400 Volt)...
The Mirage: Freq. 1 KHz, Pulswith 50µS.
The Coil: 304µH and 1,6 Ohm.
Flyback-Puls more than 450 Volt !!!

First, i lost much searchdeep and 20 Minutes later, the IRF 740 was death...

Now i used the IRF 840 with a higher Breakvoltage (500 Volt) and
all is O.K.

The IRF 9640 Breakvoltage is only - 200 Volt.

It also depends what the manufacturer means by "repetitive avalanche rated". It might still be possible to push the MOSFET out of its safe operating area, even though it doesn't get hot. There's a lot of debate as to whether a series gate resistor is necessary or not, but it would be worth adding one just to understand the problem better. Of course, the last resort would be to fit a MOSFET with a higher breakdown voltage.

[Qiaozhi]

It is not true. Frequency gives you only sweeping speed comfort, nothing more. 200Hz is enougt in most cases but 300Hz gives you comfort

I think Eric would disagree with you on this, as he often uses much higher pulse rates, sometimes as high as 10kHz. Of course, you don't want to keep the same pulse width, otherwise your battery would be flat in a few minutes. The idea is that the higher pps improves the signal-to-noise ratio of the received signal.

[kt315]

what type of FET uses whites who could know? board pics is so low res i do not see the label.
http://www.detectorprospector.com/fo...ro-whites-spp/

[daverave]

It also depends what the manufacturer means by "repetitive avalanche rated". It might still be possible to push the MOSFET out of its safe operating area, even though it doesn't get hot. There's a lot of debate as to whether a series gate resistor is necessary or not, but it would be worth adding one just to understand the problem better. Of course, the last resort would be to fit a MOSFET with a higher breakdown voltage.

Hi George..ive ordered a 500v
p-channel mosfet and i will try this but the downside is that the on resistance of the mosfet is 6 ohm and my coil is around 2.6 ohm which together i feel maybe a bit high !!!

[daverave]

I think Eric would disagree with you on this, as he often uses much higher pulse rates, sometimes as high as 10kHz. Of course, you don't want to keep the same pulse width, otherwise your battery would be flat in a few minutes. The idea is that the higher pps improves the signal-to-noise ratio of the received signal.

As far that im aware the white's tdi sl draws 500 ma so i guess the tx must be around 100 uS with frequency around 3.3 khz and the pro formance from what ive read is deep on low conductive items and even deeper with the pro version....so the higher frequency must play a part in the depth proformance which as you said the signal to noise ratio is improved at higher frequency so we can hear those deep faint signals better.

[Jose]

I have a PI, which works similarly to the one described by daverave, 3.2 Kz 100 us.
But powered only with a 12-volt gel battery.
Adenas, has an R of 1 ohms and a capacitor of 4700 mF in the drain, this limits the peak current.
Also a serial diode with the coil, which somewhat decreases the total voltage, somehow limiting the current.
The mosfet, it works with high temperature, reason why it is unthinkable for my PI, tensions greater of 12 Volts.
Comparing this PI, which works with the mentioned features, the performance is better than others, working with low frequency for example with 550Hz.
Jose

[dantech]

when i was running detector at 70 uS tx/3.2khz freq/15v supply/2.6 ohm coil...i was drawing from the supply about 300 mAH.

That is the proof that majority of battery power goes as a heat into the air. Like I said the coil can store limited energy and the rest is lost.
So there is no sense to feed excessive power to the coil. The coil is not able to radiate that power and you see it at a heat.

On my movies you can see gold and silver finger ring

[Qiaozhi]

Hi George..ive ordered a 500v
p-channel mosfet and i will try this but the downside is that the on resistance of the mosfet is 6 ohm and my coil is around 2.6 ohm which together i feel maybe a bit high !!!

With an Rds of 6R0 + coil resistance of 2R6, this gives a maximum coil current of . So you should have a good chance of allowing the current to flat-top. The best pulse rate depends on what targets you're hoping to find.

A US nickel (for example) has a of about 10us, so you would need a TX-on pulse width of to flat-top.
To achieve an average coil current of 340mA would require a TX period of
Which is a TX pulse rate of .

I know you're in the UK, so the US Nickel is just to give you the idea.

In your particular case, with a TX pulse rate of 3200pps, we have:







Hence:



Which would allow you to focus on targets with a of and below.

However, we also need to consider the coil and MOSFET parameters:







Then the during TX-on can be calculated as:



In this case, it is a waste of battery power to have a TX-on time of more than if you're happy to focus on targets with a below .

Just some food for thought.

You might want to refer here to see why flat-topping can be important ->
http://www.geotech1.com/forums/showt...top#post221739

[Ferric Toes]

There are many points in this thread which results in some confusion. I have used the IRF9640 quite a lot, but mainly in industrial PI's. i.e fixed mains operated units with rectangular noise cancelling search coils mounted under conveyor belts. It proved to be a reliable device and operated fine at a high pulse rate. For hobby detectors I generally use N channel Mosfets such as IRF740, which also is used in the TDi. Both these Mosfets should be on a heatsink that keeps their operating temperature to no more than 50degC. Mosfets have a positive temperature coefficient, so the On resistance goes up when they get hot and pulse current will drop. This is safer than silicon bipolar devices, where the resistance drops and you get thermal runaway.

I have had problems in the past when driving Mosfets direct from 555 timers. In one design the timer would fail, causing the Mosfet to turn on continually, overheat, and then usually fail as well. Never really found the cause and as proof of Murphy's Law, it was always the detectors that were sold abroad, or at the far end of the UK that failed. In the workshop the fault almost never occurred. I suspected that parasitic capacitance passing the high voltage spike back to the the 555 was the problem. Now, I always use a complementary emitter follower to drive the Mosfet gate. This both charges and discharges the gate capacitance quickly and acts as buffer between whatever the source of the drive pulse and the MosFet.

It is important to look at the coil current on a scope by inserting a 0.1 ohm resistor in the ground end of the coil. It should be a smooth exponential rise to where it levels off and the current is just that determined by the coil resistance. With short pulses it won't, of course, reach the resistive level. If there are kinks in the coil current waveform, or it starts to droop toward the end then something is not right. The gate drive must turn the Mosfet fully on at all levels of current. If it is inadequate then the Mosfet will overheat.

Regarding the use of high pulse frequencies and low pulse currents, I will tend to go this route because it is easier to sample earlier when switching off a low currrent, and which doesn't involve going into the avalanche mode of the Mosfet. The loss of signal caused by the low pulse current is made up by repeating the pulse more frequently as well as being able to sample earlier. In the development of the TDi, I made a low pulse current version which used an IRFD210 with no heatsink, running at 10,000 pps which was only slightly down on performance to the standard version. Both were set to 10uS delay as the limitation to running a shorter delay was the coil/cable characteristics.

Eric.

[daverave]

Hi Eric...i found what you said about low pulse current made up by higher frequency very interesting which ive never really realized...when you say about low pulse current i guess you mean like 100 us tx pulsed at higher frequency as in tdi around 3.3 khz.

[daverave]

With an Rds of 6R0 + coil resistance of 2R6, this gives a maximum coil current of . So you should have a good chance of allowing the current to flat-top. The best pulse rate depends on what targets you're hoping to find.

A US nickel (for example) has a of about 10us, so you would need a TX-on pulse width of to flat-top.
To achieve an average coil current of 340mA would require a TX period of
Which is a TX pulse rate of .

I know you're in the UK, so the US Nickel is just to give you the idea.

In your particular case, with a TX pulse rate of 3200pps, we have:







Hence:



Which would allow you to focus on targets with a of and below.

However, we also need to consider the coil and MOSFET parameters:







Then the during TX-on can be calculated as:



In this case, it is a waste of battery power to have a TX-on time of more than if you're happy to focus on targets with a below .

Just some food for thought.

You might want to refer here to see why flat-topping can be important ->
http://www.geotech1.com/forums/showt...top#post221739

Hi George
thanks for your interesting formulas which i will take on board.......i do have a US nickle which i find the best coin to test a detector for low conductive depth......i guess if i use the 500v 6ohm mosfet then my detector will only really be good for very low conductors like nickles or below.

[Qiaozhi]

I have had problems in the past when driving Mosfets direct from 555 timers. In one design the timer would fail, causing the Mosfet to turn on continually, overheat, and then usually fail as well. Never really found the cause and as proof of Murphy's Law, it was always the detectors that were sold abroad, or at the far end of the UK that failed. In the workshop the fault almost never occurred. I suspected that parasitic capacitance passing the high voltage spike back to the the 555 was the problem. Now, I always use a complementary emitter follower to drive the Mosfet gate. This both charges and discharges the gate capacitance quickly and acts as buffer between whatever the source of the drive pulse and the MosFet.

Yes, I suspect Dave's problem is very similar. It must have something to do with the MOSFET being driven directly by the 555 timer. Of course, the original Surf-PI didn't use a MOSFET, so the situation would have been different, otherwise we have heard of many Surf-PIs dying in the field, and that's definitely not the case.

Regarding the use of high pulse frequencies and low pulse currents, I will tend to go this route because it is easier to sample earlier when switching off a low currrent, and which doesn't involve going into the avalanche mode of the Mosfet. The loss of signal caused by the low pulse current is made up by repeating the pulse more frequently as well as being able to sample earlier. In the development of the TDi, I made a low pulse current version which used an IRFD210 with no heatsink, running at 10,000 pps which was only slightly down on performance to the standard version. Both were set to 10uS delay as the limitation to running a shorter delay was the coil/cable characteristics.

Eric.

I did some experiments recently where I tried various TX pulse rates and TX-on times, and can confirm that there's very little discrepancy in performance between using low and high pulse rates. But one benefit is being able to sample earlier at higher pulse rates, which also enables sampling higher up the curve at the preamp output. Hence the DC offset is lower, and you can use extra amplification in the following stages to improve performance.