I also bought an Aneng AN8008 and pretty happy with it!!!
I was wondering if matching the resistor will really give us better results...
As long as the resistors are good enouth like 1% and do not drift, the self-test should take into account the small differences in the resistors and also in the ADCs of the atmega. What should be critical is just the Voltage regulator/reference.
M'I missing something? Does the matched resistors really improve precision and or accuracy or something else?
Mauro
I'm definitely a fan of this little 9999 count meter. I'd like to understand more about it but the thread on here is huge already and I don't want to get too distracted. I'm very tempted to get another AN8008 and put it in a bench enclosure though. I've already burned through 4 batteries by using the continuity and diode checker quite a bit.
The post where Marcus replied to my questions about the resistors and tolerances
is here. If I understand it correctly, the probing resistors are extremely important because the software has 1 value attributed to each of the two
SETS of probing resistors. This means the device can't calibrate each individual probing resistor value and compensate for the difference between channels. All calculations are based around the idea that 1mA read across test point 1 can and will be 1mA if, for example you reversed the device under test and did the same test sequence again. Of course we are interested in resolution readings that are much smaller than 1mA in resolution.
Again, IF I understand what Marcus said in the message linked above, the accuracy of the resistor value is not terribly important, the matching tolerance is the key. Basically we can pay for an extremely accurate set of resistors at 0.1% and we end up getting the tolerance we needed. The tolerance of these resistors is a byproduct of their accuracy. If the accuracy of the resistors were important, Marcus would have had a different answer to my question about temperature coefficients. -That bit of info and his reply to my question about 500ohm and 500k resistor parallel combos is where I'm basing my understanding.
The biggest factor beginners need to understand is the difference between accuracy and tolerance. Accuracy is how close a part is to it's designated value. The only way to know accuracy is with metrology grade equipment or by paying a calibration lab to compare your junk to their super expensive stuff and tell you how wrong you are on a piece of paper and a sticker. Determining tolerance is much easier. Tolerance is determined by resolution based measurements. That one took me awhile to really understand. Hopefully it helps the next person.
If you look at the measured specs of the resistors I posted before, the 680 1% resistors from my AY-AT clone are at 0.82% total. That's a good bit higher than 0.1% specified in the project and it is a random sample of 1 kit that in no way indicates how good or bad others' devices may be.
If we are not worried about accuracy and understand that tolerance is key, things change. If we have the ability to measure resistors and compare tolerances at a higher resolution with repeatable results we can create our own high tolerance set. The proper way to do this would probably be with a 5.5 digit or greater resolution meter where the resolution is much better defined.
3.5 digits - 1,999 counts
3.75 digits - 3,999 counts
4.0 digits - 9,999 counts (AN8008)
4.5 digits - 19,999 counts
5.5 digits - 199,999 counts
6.5 digits - 1,999,999 counts
If I base my measurements on a meter with a tolerance of +3 counts that means I can't trust that last digit to be repeatable. So basically my metrics I posted are total junk to the pros watching my tomfoolery. However my junk measurements have been repeatable in scale on 3 different occasions where I have measured them. The final readings may vary but the same groups of resistors are close in value, and my matched 5% resistors consistently perform far closer in total tolerence than the ones from the kit. Also the 0.1% resistors are the most consistent on multiple measurement sessions. They also show the relative accuracy of the meter as they are probably much more accurate than the meter. I'll be doing more testing soon and will show much more in a video within the next week or so. I'm filming stuff now and have the main schematic version of the tester on a breadboard.
-Jake
Home
Help
Search
Login
Register
Topic: $20 LCR ESR Transistor checker project (Read 4240153 times)