@graham-xrf attached is the footprint I made for an LCC10 for the PIN diode. You will need to rename from LCC10.txt to LCC10.kicad_mod and place it in an appropriate place. It has not been rigorously checked, but I hope it is close!
-
Welcome back Guest! Did you know you can mentor other members here at H-M? If not, please check out our Relaunch of Hobby Machinist Mentoring Program!
Needing more than a spark test?
- Thread starter graham-xrf
- Start date
Hi Bruce
Thanks for the diode footprint.
----------------------------------------------------
I made a symbol for the AD7622.
So far as I can tell, it is identical to the AD7667 except for the name of Pin39, which changes from IN- to INGND
The one I made looks like this..
You or I could edit a copy, changing the Pin39 name. We need to check that the pin 39 is the only pin with a different name.
The footprint layout is one of the standard ones for 48 pin quad flatpack I found in KiCad.
It is Package_QFP:LQFP-48-1EP_7x7mm_P0.5mm_EP3.6x3.6mm_ThermalVias
That is the one I assigned in the symbol properties. It should be immediately available to you.
Hang in there - just having a look at exporting.
Thanks for the diode footprint.
----------------------------------------------------
I made a symbol for the AD7622.
So far as I can tell, it is identical to the AD7667 except for the name of Pin39, which changes from IN- to INGND
The one I made looks like this..
You or I could edit a copy, changing the Pin39 name. We need to check that the pin 39 is the only pin with a different name.
The footprint layout is one of the standard ones for 48 pin quad flatpack I found in KiCad.
It is Package_QFP:LQFP-48-1EP_7x7mm_P0.5mm_EP3.6x3.6mm_ThermalVias
That is the one I assigned in the symbol properties. It should be immediately available to you.
Hang in there - just having a look at exporting.
Updated and even messier layout. You can see the guard ring on U2. I don't know how to split C2, since the guard ring is in place. I could always do an air bridge, but I'd like to avoid that! Solder each 2pF cap to the pad with the other in the air, and a wire across the two pads in the air.
Found a symbol and footprint for the AD7667ASTZRL on snapeda. https://www.snapeda.com/search/?q=AD7667ASTZRL&search-type=parts
Found a symbol and footprint for the AD7667ASTZRL on snapeda. https://www.snapeda.com/search/?q=AD7667ASTZRL&search-type=parts
OK - too many things to get into at once here.
Firstly, about C2. Easy enough. It just becomes two capacitors in series, in a straight line, so that the outer ends do not come near each other.
One end of the capacitor that connects to the inverting input "drops into" the guard ring region, as does one end of the feedback resistor. The guard ring track goes right under both of them.
The "other" capacitor of the series pair goes to the opamp output, using the same track that the feedback resistor is on.
----------------------------------------------------
Regarding the diode bias connection circuit.
Yes - I do prefer yours, but now I have to look hard at biasing circuits.
(Compare to the connection used in the Pocket Geiger. Direct coupled at that point, and the dark current forced to run into theTIA, and produce a big offset, then forcing AC coupling after the TIA)
Your circuit shows the photodiode having to deliver it's little pulse into a relatively low 100K impedance, in parallel with the route via C1 into the TIA.
The bias dark current is 4nA reverse biased. For (say) 10V bias, that would supply even if the resistor was 2.5GΩ. There is plenty of room to make the bias resistor 1MΩ, or 10MΩ. Maybe two 1MΩ resistors with a decoupling (anti-noise) capacitor at the join.
Keep in mind that internally, the reversed biased diode is shunted with the equivalent of 40MΩ.
BUT..
Internally, the carriers generated by the photons connect to the terminals though the very much more conductive P and N diode bulk resistance stuff. I still think that, given we have the very high impedance input to the TIA, it's a bit of a waste to let any more than necessary get lost via the relatively lower impedance of R1
Let me have a go at the symbol library first, for making the AD7667 symbol.
Firstly, about C2. Easy enough. It just becomes two capacitors in series, in a straight line, so that the outer ends do not come near each other.
One end of the capacitor that connects to the inverting input "drops into" the guard ring region, as does one end of the feedback resistor. The guard ring track goes right under both of them.
The "other" capacitor of the series pair goes to the opamp output, using the same track that the feedback resistor is on.
----------------------------------------------------
Regarding the diode bias connection circuit.
Yes - I do prefer yours, but now I have to look hard at biasing circuits.
(Compare to the connection used in the Pocket Geiger. Direct coupled at that point, and the dark current forced to run into theTIA, and produce a big offset, then forcing AC coupling after the TIA)
Your circuit shows the photodiode having to deliver it's little pulse into a relatively low 100K impedance, in parallel with the route via C1 into the TIA.
The bias dark current is 4nA reverse biased. For (say) 10V bias, that would supply even if the resistor was 2.5GΩ. There is plenty of room to make the bias resistor 1MΩ, or 10MΩ. Maybe two 1MΩ resistors with a decoupling (anti-noise) capacitor at the join.
Keep in mind that internally, the reversed biased diode is shunted with the equivalent of 40MΩ.
BUT..
Internally, the carriers generated by the photons connect to the terminals though the very much more conductive P and N diode bulk resistance stuff. I still think that, given we have the very high impedance input to the TIA, it's a bit of a waste to let any more than necessary get lost via the relatively lower impedance of R1
Let me have a go at the symbol library first, for making the AD7667 symbol.
Last edited:
I can easily split C2, with the "inner" C being split by the guard ring and the outer C outside the guard ring. R1, well that is outside the dense zones, it is just a value! Easy to split, increase and filter. Can do that in a couple of minutes.
Finally got through the layout of this so far. Had the DRC complain about the footprint of the part that was in its own libraries. Had to allow reduced via sizes and those faults went away. I have to add in the supply plane split on one of the inner layers. If I add the ADC and the Teensy, it will be interesting to say the least. Probably have to add a pair of layers.
Since I haven't done many boards in KiCAD, I have to figure out where/how one sets the board size! Or more accurately figure it out again. Did it once, and apparently I don't remember how. You would think this would be a bit easier. If I was using the tool more often, I'd remember it. Off to the search engine...
Without the split C2 layout. Everything is connected, except the ADC, power planes, and the Teensy and display. Had to use a reduced size via near U4, standard one was too big to fit. DRC is complaining about my footprint for D1. Apparently the solder mask is cutting through the silkscreen. As a rank beginner - I just didn't know...
For @RJSakowski
I received two thorium lamp mantles (eBay) in one of those 150 x 120 plastic flat envelopes with label.
The mantles inside were in their own plastic bags, and put between a double layer of paper around them, I suppose to keep them flat.
I just assume that the big racket of alpha particles did not penetrate enough to escape.
I have seen in one paper online, the writers used 3 x 3.4mm of "Plexiglass" to stop any "beta", so they could measure the X-rays and gamma coming out of thorium mantles. Beta is electrons, so I would have thought that some aluminium foil would stop them. I did not have any "Plexiglass". I am not sure what the difference might be between Plexiglass and polycarbonate sheet. One is acryllic I suppose.
I am surprised that this little envelope did not get detected and stopped in the post. That said, "Royal Mail" sold to private, are hardly great anymore. I would have called it "somewhat radioactive"!
The first 588cpm peak is with the counter above the envelope, maybe 5mm spaced. The instrument dosage alarm went off well below the 120cpm
The next peak is with 30 microns thick aluminium cooking foil over it. There seems hardly any difference. I know it shows about 460cpm, but if held slightly differently, the count can go over 750, or 1000.
The last peak to the right hitting 120cpm is when the mantles are under 12.5mm of plate glass (my A4 size on-desk impromptu surface plate substitute). Lacking the Plexiglass, I thought some other insulator might stop electrons, but glass is not just carbon and hydrogen and oxygen. It has silicon, maybe lead additves, and all the stuff that goes into glass.
Hence I ask - through the foil, was I looking at mostly X-rays without beta?
I received two thorium lamp mantles (eBay) in one of those 150 x 120 plastic flat envelopes with label.
The mantles inside were in their own plastic bags, and put between a double layer of paper around them, I suppose to keep them flat.
I just assume that the big racket of alpha particles did not penetrate enough to escape.
I have seen in one paper online, the writers used 3 x 3.4mm of "Plexiglass" to stop any "beta", so they could measure the X-rays and gamma coming out of thorium mantles. Beta is electrons, so I would have thought that some aluminium foil would stop them. I did not have any "Plexiglass". I am not sure what the difference might be between Plexiglass and polycarbonate sheet. One is acryllic I suppose.
I am surprised that this little envelope did not get detected and stopped in the post. That said, "Royal Mail" sold to private, are hardly great anymore. I would have called it "somewhat radioactive"!
The first 588cpm peak is with the counter above the envelope, maybe 5mm spaced. The instrument dosage alarm went off well below the 120cpm
The next peak is with 30 microns thick aluminium cooking foil over it. There seems hardly any difference. I know it shows about 460cpm, but if held slightly differently, the count can go over 750, or 1000.
The last peak to the right hitting 120cpm is when the mantles are under 12.5mm of plate glass (my A4 size on-desk impromptu surface plate substitute). Lacking the Plexiglass, I thought some other insulator might stop electrons, but glass is not just carbon and hydrogen and oxygen. It has silicon, maybe lead additves, and all the stuff that goes into glass.
Hence I ask - through the foil, was I looking at mostly X-rays without beta?
Regardless you are still learning KiCad, you are storming along!
OK - I now have KiCad open. I will try for your edited AD7667
I will address all the various bias stuff soon. There are circuits where you can have arbitrary bias, be direct coupled, and apparently not have a dark current offset that sends the output off the scale. I have to get to understand all these fully. One example is shown here.
The distance traveled by a beta particle will depend upon the energy of the beta particle and the average atomic number of the attenuator. I would expect that it would similar to half life in that the distance would be that point where 50% of the particles are attenuated. so to effectively stop the betas, you would need to use multiples of the thickness. Here is a reference. http://www.phys.utk.edu/labs/modphys/AttenuationRadiation.pdf Note that aluminum in the thickness of cooking foil won't be too effective. You would probably want about 1mm.
Plexiglas is known as Perspex in the UK but I would expect that polycarbonate would behave similarly. Plate glass isn't likely to contain lead. Lead glass is used for crystal and as a shield for CRT's. It would be soda lime glass so mostly SiO2 with CaO and Na2CO3. https://www.britannica.com/technology/glass
I wouldn't call it storming... More like plodding. Having difficulties setting up power planes and seeing them. It's frustrating when one has done it before, and can't quite pull it off a second time. Somewhere there is a tutorial on this, but haven't found it yet.
The DRC is going crazy on some traces, which look ok. I can fix some of them, but when I rerun the DRC, the errors seem to return. They might be related to grid size stuff, but the traces are showing (when I do them) that they are connected. Primary errors are a trace not connected to a pad, or the via. The DFN substrate is connected to V-, it comes up as a pin, and it is insisting on being connected. I was planning to have a via to a power plane, but I haven't set it up yet.
Split cap is in, added increased resistance and cap for the bias circuit.
Since your TIA's Rf is only 240K I think your guard band efforts probably are overkill. You also may not need an opamp with a 4GHz GBW. While it's a little more money the LT1028 has lower voltage noise and still offers a 50MHz GBW. Its low frequency noise is much better, too.
I had looked at the LT1028 for experimenting with Earth-field NMR but that project has been on hold for awhile.