Needing more than a spark test?

[homebrewed]

In these things, if you want to use the fastest it can go, (warp mode), there is a minimum time between conversions of 1mS, should you be wanting to go the slowest. If course, that is real slow, but lets you get up to 1MSPS when you go fast. Choosing "normal" mode lets it run at 800kSPS maximum, but should you desire, you can clock it along as slow as you like (I think).

Forgive that I don't know the detail of how you are having the Teensy talk to the AD7667, but I did not think you were constrained to having the Teensy be stuck waiting for the ADC when the event rate is 1.25uS.

I know this sounds simplistic, but can the Teensy initiate a conversion by setting /CNVST to LOW, and then go off and do whatever else it likes in it's loop until 1.25 (or more) microseconds later, and then check the pin 29 BUSY signal for data_ready?

I get it that the available delay in the Teensy may not be suitable. There are likely a whole lot of different solutions to this. Last I had a (almost) similar scene with a 32-bit microcontroller, I resorted to a little interrupt routine to decrement my own delay. The main housekeeping loop did not pause in any noticeable way.

I take it that the Teensy has to take out 8 bits at a time.

Fortunately, the driver program doesn't have to wait for the ADC to perform its conversion. If /CNVST _and_ /RD are both driven low, the ADC places the results of the PREVIOUS conversion on its data pins and also initiates the _next_ conversion. The impact of noise is minimized because the conversion starts with the MSB first, so by the time the ADC's SAR is working on the LSBs the data-read is done. The delays I have to deal with are all in the 10-45nS range, aside from the somewhat slower GPIO access time, so very short compared to the 1us conversion time. That's per the data sheet anyway.

My ADC board includes a jumper that shorts /CNVS and /RD so this mode of operation requires one less connection between the Teensy and ADC board. Right now I also am not using the BUSY signal. Hopefully that won't be necessary.

The Teensy4.0 has enough of GPIO6's bits brought out to perform one 8-bit read or write in parallel, although there is some bit-shifting needed to get 8 bits. Specifically, GPIO6 is a 32 bit register and the useful bits are selected by ANDING 0x0f0f0000 with the contents of GPIO6_DR then shifting and ORING to turn those separate 4-bit nybbles into an 8 bit quantity. To read 16 bits two successive 8-bit reads have to be done (but the bit shifts are different to occupy the upper 8 bits of my data word). That's where the ADC's BYTESWAP pin comes in.

It's surprising that it takes about 4X longer for the data pins to settle after changing BYTESWAP, compared to setting /RD low, but that's how it is.

The Teensy4.1 has 16 contiguous bits brought out so it definitely is better suited for this. Just one read/shift operation and no messing with the BYTESWAP pin. It still has to wait 10nS for the ADC's data pins to settle.

It's possible that the slower peripheral clock has enough delay that it may not be necessary to use as many wait states as I currently have put in my ISR code. Something TBD.

 

[graham-xrf]

60 MHz stuff, just isn't that bad... Sometimes all that is required is a small series resistor at the source. Lots of ways to deal with this, including controlled impedance lines.

Yes - 60MHz is not so bad. That is, so long as it stays on the PCB. It's real hard to make it go any distance without it needing a proper transmission line. If you want to probe to display it, use a home-made resistors 20:1 probe with a 50Ω cable into the 50Ω scope input.

It's not so much about the frequency. It is about the speed of the edge transition, and whether that has to meet up with other clocks in the chip with low jitter. This time, I will be scrutinizing the pulse rise and fall transition times specifications somewhat closely!
-------------------------------------
Right now, the bit that has to get attention is the FreeCAD model, and my lead (Pb) construction. My past attempt used link groups to hold several bits to move together. I am getting up to speed on using Assembly 4. The program is a powerful thing, but I don't need a whole lot of it's features.

I do insist that the photodiode die please stay with, and be moved about, along with its own ceramic package, and the little bit of PCB. This is coming on OK. I may get up to a trial minor heat-up of a small piece of exemplary lead. Depending on how it goes, I may just keep adding more, or I may have to adjust the heating/melting arrangements.

 

[graham-xrf]

To read 16 bits two successive 8-bit reads have to be done (but the bit shifts are different to occupy the upper 8 bits of my data word). That's where the ADC's BYTESWAP pin comes in.

Hi Mark
Just so I know, if one were considering using the Teensy SPI Library, and take data out serially, is it still only 8 bits at a time?

 

[WobblyHand]

For the moment, I am going to use some 9V alkaline batteries in series for the PIN bias voltage. Probably start out with 3 of them. Had to order some of those clip on terminals, thought I had some, but apparently no longer.

Back to the Pocket Geiger. I removed the LT1615 switcher. Powered the board via a 12V (switcher). Since I haven't reverse biased D3, the 12V (with some diode loss) is being put on the PIN diode. It may be attenuated, but the first stage is amplifying it, and so is the second. The result is that the output is all over the place in the absence of signal. To make things more fun, if I touch the PIN diode face, I get a full saturated square wave output. Touching the back of the PCB does not saturate the output of the analog stages. It would seem this supply is too noisy - not surprising, it is a multi-voltage wall wart. So need to locate a 12V battery. Or a 12V linear supply.

Way back in the depths of my garage is a linear 12VDC supply, out to fetch it. As I remembered, it is an old fashioned linear supply, with ancient TO-3 pass devices. I'm sure it is not high performance, but it should not have switching noise. Have to wait for it to warm up, is was cold soaked at 14F in the garage last night. I will hook it up and see if the Pocket Geiger output is a bit more stable.

Not really. I am getting some 4 KHz oscillation out of the second stage. Time for a 12V battery...

 

[homebrewed]

For the moment, I am going to use some 9V alkaline batteries in series for the PIN bias voltage. Probably start out with 3 of them. Had to order some of those clip on terminals, thought I had some, but apparently no longer.

Back to the Pocket Geiger. I removed the LT1615 switcher. Powered the board via a 12V (switcher). Since I haven't reverse biased D3, the 12V (with some diode loss) is being put on the PIN diode. It may be attenuated, but the first stage is amplifying it, and so is the second. The result is that the output is all over the place in the absence of signal. To make things more fun, if I touch the PIN diode face, I get a full saturated square wave output. Touching the back of the PCB does not saturate the output of the analog stages. It would seem this supply is too noisy - not surprising, it is a multi-voltage wall wart. So need to locate a 12V battery. Or a 12V linear supply.

Way back in the depths of my garage is a linear 12VDC supply, out to fetch it. As I remembered, it is an old fashioned linear supply, with ancient TO-3 pass devices. I'm sure it is not high performance, but it should not have switching noise. Have to wait for it to warm up, is was cold soaked at 14F in the garage last night. I will hook it up and see if the Pocket Geiger output is a bit more stable.

Not really. I am getting some 4 KHz oscillation out of the second stage. Time for a 12V battery...

The PIN diode is a huge capacitor in addition to an xray detector so there's the possibility of feedback due to the capacitive coupling + high overall circuit gain. Moving the copper foil over to wrap it around the electronics should help prevent oscillations. Of course you want to make sure it doesn't short something that you DON'T want shorted to ground

Aluminum foil over the diode wouldn't attenuate x-rays much but now that I've been burned by that annoying aluminum peak I'm trying to eliminate aluminum where it makes sense to do so. I hadn't originally thought that the ~6Kev xrays from iron would get the aluminum going so much but that turned out to be an incorrect assumption!

BTW I have a really old dual 0-40V linear bench supply I'm using for this project. One output provides the bias voltage (albeit filtered by a 2-pole RC filter network in addition to the pocket geiger's RC filter) and the other provides -10V for the signal conditioning board. I have a separate single-output linear supply for the +10 that runs the pocket geiger and signal conditioning boards.

 

[WobblyHand]

It think I may need to line isolate my scope I powered the board with a 12V battery and I still see 60 Hz stuff plus some high frequency trash. The battery is a SLA 12V, that can't supply too much current - it's more or less used up/dead, but I did measure over 12V even under load of the board. I'm seeing typical issues with scopes and somewhat cheap probes - ground connections that are not so hot, etc. Don't remember it being this bad before, but it has been awhile.

First I will check the digital outputs - if they are moving about, the comparator is coming out. And there's no activity on the comparators.

The diode is very sensitive to a metallic object near the face. The closer you get the more it misbehaves. How this was ever set up and stabilized seems amazing. Putting it the foil back makes the circuit difficult to test. Got to come up with something better.

After getting frustrated with that, I decided to open an old smoke detector and liberate the source. This is what I found after getting it all apart.

Not that it helps me at this point. My detector circuit (Pocket Geiger) isn't working, at least in my estimation. Is the first photo, inside the peening, the actual source?

 

[graham-xrf]

For the moment, I am going to use some 9V alkaline batteries in series for the PIN bias voltage. Probably start out with 3 of them. Had to order some of those clip on terminals, thought I had some, but apparently no longer.

You don't need -27V, You don't need -18V. You barely need the 9V from one battery!
The dark current, and its noise, and the reverse bias capacitance, and it's effect on the signal are very closely specified - with plots even.
Choose where you want it to be!

Back to the Pocket Geiger. I removed the LT1615 switcher. Powered the board via a 12V (switcher). Since I haven't reverse biased D3, the 12V (with some diode loss) is being put on the PIN diode. It may be attenuated, but the first stage is amplifying it, and so is the second. The result is that the output is all over the place in the absence of signal.

Hardly surprising!
If you go to the trouble of hot air removing the LT1651 switcher, (transposing some part number) digits , then why replace the voltage with one from another switcher? At this stage, even a 1.5V AAA would do, or no bias at all!

To make things more fun, if I touch the PIN diode face, I get a full saturated square wave output. Touching the back of the PCB does not saturate the output of the analog stages. It would seem this supply is too noisy - not surprising, it is a multi-voltage wall wart. So need to locate a 12V battery. Or a 12V linear supply.

Probably just about any supply to any part of the circuit upstream of the first opamp will be hell to filter clean enough. Putting finger on the diode? Good Lord, it's a stage which arrives attempting to deliver a gain of 66 million! Even though it can't do that, it was always going to bang it's rails, even from waving a hand too close.

Way back in the depths of my garage is a linear 12VDC supply, out to fetch it. As I remembered, it is an old fashioned linear supply, with ancient TO-3 pass devices. I'm sure it is not high performance, but it should not have switching noise. Have to wait for it to warm up, is was cold soaked at 14F in the garage last night. I will hook it up and see if the Pocket Geiger output is a bit more stable.

Not really. I am getting some 4 KHz oscillation out of the second stage. Time for a 12V battery...

If you just used the 9V battery to test it, you would save a delve through the garage. That said, delving through the garage, and coming across stuff I have needed, but couldn't find at the time, is exactly what I end up doing all the time.

I accept that my route, which is cut the diode with a little PCB it is soldered to off, and bin the remainder, runs counter to every instinct of anyone who has forked out the $$ for something which has just arrived in the post, but that is a personality trait fail that I have. In my view, except for acquiring the diode, nothing about the Pocket Geiger, and its circuit was of any further interest.

 

[graham-xrf]

It think I may need to line isolate my scope I powered the board with a 12V battery and I still see 60 Hz stuff plus some high frequency trash. The battery is a SLA 12V, that can't supply too much current - it's more or less used up/dead, but I did measure over 12V even under load of the board. I'm seeing typical issues with scopes and somewhat cheap probes - ground connections that are not so hot, etc. Don't remember it being this bad before, but it has been awhile.

If your scope is a 2-channel type, normally one can use two probes, with one set to be inverted, and summed. Go through the probe setup and square wave nulling procedure. I am thinking you already know all this. Anyway, you can use one probe on the 0V and the other to where you want to probe, knowing the interference will cancel. The 60Hz stuff, and possibly quite a lot of "high frequency trash" will disappear.

First I will check the digital outputs - if they are moving about, the comparator is coming out. And there's no activity on the comparators.
The diode is very sensitive to a metallic object near the face. The closer you get the more it misbehaves. How this was ever set up and stabilized seems amazing. Putting it the foil back makes the circuit difficult to test. Got to come up with something better.

We did say, way back, that all of this would happen. Giving a big fat flat huge area diode a piece of external metal to participate in being a capacitor, yet coupled to the outside fields would do this. It's not unlike those elevator capacitive sensing pushbuttons.

After getting frustrated with that, I decided to open an old smoke detector and liberate the source. This is what I found after getting it all apart.
View attachment 435843View attachment 435844View attachment 435845
Not that it helps me at this point. My detector circuit (Pocket Geiger) isn't working, at least in my estimation. Is the first photo, inside the peening, the actual source?

I know that the exact mechanical detail of the contact metal shape may be a bit different, but the middle bit should be very similar. I managed to liberate it from the contact ring just by viciously bending the metal away from it. As I recall, I started filing across one peening with a little watchmaker's warding file, but then I just went ahead with and bent it away with pliers. I now have the little 1/4" button free of the contact metal.

 

[WobblyHand]

You don't need -27V, You don't need -18V. You barely need the 9V from one battery!
The dark current, and its noise, and the reverse bias capacitance, and it's effect on the signal are very closely specified - with plots even.
Choose where you want it to be!

Hardly surprising!
If you go to the trouble of hot air removing the LT1651 switcher, (transposing some part number) digits , then why replace the voltage with one from another switcher? At this stage, even a 1.5V AAA would do, or no bias at all!

Probably just about any supply to any part of the circuit upstream of the first opamp will be hell to filter clean enough. Putting finger on the diode? Good Lord, it's a stage which arrives attempting to deliver a gain of 66 million! Even though it can't do that, it was always going to bang it's rails, even from waving a hand too close.


If you just used the 9V battery to test it, you would save a delve through the garage. That said, delving through the garage, and coming across stuff I have needed, but couldn't find at the time, is exactly what I end up doing all the time.

I accept that my route, which is cut the diode with a little PCB it is soldered to off, and bin the remainder, runs counter to every instinct of anyone who has forked out the $$ for something which has just arrived in the post, but that is a personality trait fail that I have. In my view, except for acquiring the diode, nothing about the Pocket Geiger, and its circuit was of any further interest.

At this point, I am inclined to agree with you.

The current circuit is awkward to say the least. It's not that easy to simply insert the bias with a simple battery, at least not with further surgery. The existing circuit has one side of the PIN diode connected to the op amp which is sitting at 2.9V not 0V. I would need to remove a resistor or diode to properly add in the battery.

Considering that the circuit isn't stable in the open air, it's hard for me to think that I want to spend much more time with it. Wrapping the unit with a grounded foil, and some insulating plastic is just adding capacitance to the circuit. Seems its a miracle that it does anything useful at all.

It would be nice to remove the diode from the PCB and start afresh. I can see why you want to saw it off. I thought, maybe there is some redeeming value in the circuit, let's try it out. Well I did try it.

I have concluded, it's not a very good basis to work with. And yeah touching the diode was not ever going to be good. I was expecting railing from that. There's no way that op amp can deliver the gain of 66Meg. The op amp has to be hitting some over riding limit, like slew rate or something like that. Back to the books, and simulations. Guess I will have to attack why I had issues with LTspice...

 

[WobblyHand]

I can try the differential signal approach on my scope. I've done that before. I will put off a resistor to isolate the diode bias - that ought to help some.

Had hoped that the existing circuit had some merit, even as hokey as it was. But it seems its not suitable to the task. I'll come up with something or another, or hope for some guidance from this group. I'm not discouraged, since it was only a wild hope that the circuit could have some utility. Means I have to be more serious about learning more about TIA's.