Tweet
Well, i'm bit confused now. Here is the part of old Chance timing, V8 software, recorded long time ago, traces are preamp output and ADC sample command, original coil. Pic.1 is after long pulse, sampling is after some 28uS (not 18 as I thought, this is probably coil speed, I remembered wrong value). On new software adjustable down to some 12uS or somewhat less, but better re-check, my memory is not reliable. Pic.2 is sampling after burst cycle, Pic.3 is complete burst, EF sample taken before it, also before long pulse.
-
-
From the START of flyback it is 8us later when the A-D opens up for 12.5us with a GI setting of 10.
I was wrong in the number of '19' short pulses before the long pulse and flyback. It is only 15 short pulses, all 40us apart, first pulse is 5us wide, fifteenth pulse is 16us wide, before the 120us long pulse and then flyback. Tepco says that in an early software version, the coil is shorted between short pulses by the second mosfet in order to store energy in the coil until it is released after the long pulse of 120us. The short end of pulse 'flyback' I am seeing is probably the coil shorting interval between the short pulses.Comment
-
Yes the triggering is a bit tricky on this using the diodes as a signal/trigger point. It is a very complex TX signal with a lot going on. DId you trigger on the gate drive for the TX mosfet? As I am trying to see both the flyback and the relationship to it of the A-D sample command I have to use Channel 1 for the A-D and the other for receive signal & trigger unless I go external trigger on my time base to the 'TO' control lead of the uC. Is that what you did?Originally posted by Tepco View PostComment
-
Yes, i'm wrong again, 12.5uS is sample width, not delay. So, to conclude, at guard interval 10 delay is some 8uS (probably 9 ), at GI 30 around 27uS. Processor is clocked at 11.0592MHz, so cycle time is around 0.9uS, this is probably GI step. Not surprising that GI 30 performance will be very poor, and coil capable for GI 10 is very fast indeed.
I triggered scope from MOSFET gate, but DSO can provide variable delay, so I can scan entire waveform, will be bit more tricky with analog scope.Comment
-
Just ran with external trigger from mosfet gate and CHannel 2 on opamp output, Channel 1 still on the A-D pin 5. From end of long pulse ( start of flyback) it is 8us later that A-D window opens.Comment
-
So, 9 cycles are 8.1uS, try other GI settings in ,say, 10 increments, to see if 0.904us per step fits well.Comment
-
Fits beautifully! With 80 GI steps measures 71us change with my old analog scope
80 X .904 =72.32
We'll get this CHANCE thing figured out yet!Comment
-
Good to see you are getting it figured out. Tepco's 3JPG reply #91 shows 3 samples taken after fly back, curious which ones move when you adjust GI?Originally posted by baum7154 View PostFits beautifully! With 80 GI steps measures 71us change with my old analog scope 80 X .904 =72.32
We'll get this CHANCE thing figured out yet!Comment
-
Keep in mind that his screens are from version 0.8.4 several software generations ago. From what I have seen since running these tests the only one that moves with different GI inputs is the one after the long pulse. All other A-D samples appear to be fixed. There is still a lot to figure out on this thing but it is going to be fun. Thanks for your help and input on this effort.
One thing that I am curious about is where is a sample delay of 99 to be used on this detector? This is the equivalent of 88us after the end of the long pulse. 88us decay time /5= a target with a 17.5us TC...roughly. So it is moving the A-D window out 88us to exclude responses faster than that? HmmmmComment
-
Maybe that 99 Guard Interval is good for looking for underground pipes etc without being bothered by all those pesky gold nuggets
Comment
-
CORRECTION!Originally posted by baum7154 View Post
More scope time on this last night. Recounted the short pulses and there are 16 short ones followed by a 120us long pulse. At the end of the long pulse flyback begins, and the coil volts cross zero 6us later. The software is Version 1.2.1 on this detector.
Performed critical damping with the scope while listening to the detector audio and the Guard Interval set at '10' (fastest setting). When damping was optimum the audio changed from a raspy buzz to melodious tones as the detector began to operate. A quick push on the ZERO button quieted the noise. You can add a little tweak to the damping resistance if you want under damp it or alternatively do this in air. I should note that I started the damping adjustment from a resistance below the value for critical damping and this is manifested by the raspy buzz of the audio.
DanComment
-
Attached here are some pictures reposted from the CHANCE PI BUILD thread in the Tech Forum. The #1 pic of the coil flyback shows the coil decay at zero to be about 6us. With the measurement after the amp (pic #5)the zero point is much longer...time to upgrade to a dual amp configuration in order to sample faster!
Attached here are some pictures with a description of each screen shot:
1. This picture is of the coil flyback measured at the diodes on the amp input. The horizontal line at midpoint is the A-D sample window
P1060252.jpg
2. This picture is of the last of 8 short pulses and the 120us long pulse and coil flyback,again taken at the diodes. The Guard Interval moves the single A-D sample shown at the far right 0.9us for each increment of GI input. All others are unaffected by GI settings.
P1060258.jpg
3. This picture is of the 16 short pulses with their corresponding A-D dual samples in the 2nd horizontal row from the top.
P1060263.jpg
4. This picture was taken at the amp output and at pin 5 of the A-D. It shows the A-D sample at its position with a Guard Interval of 10. This shot is of the improperly damped coil and the second flyback image at the right of the trace is there because damping is not correct. The detector audio is a raspy buzz in this state . Note the increased flyback time at the amp output compared to the measurement at the input to the amp.
P1060268.jpg
5. This picture is taken at the amp out and A-D pin 5. It shows a properly damped coil and the A-D sample. Note that the image to the right from the previous shot is now gone. The detector audio is now in a quiet state.
P1060274.jpgComment
-
After making these measurements and photos I decided to make a 2 stage amp for the front end of the CHANCE PI detector. The amp is now in place in my detector and the result is that the 16us coil decay in the previous photos P1060274.jpg has been reduced to about a 7us coil decay. I'll be posting photos of the amp and the results in the CHANCE PI thread of the TECH FORUM soon.Comment
-
Baum7154,Originally posted by baum7154 View Post
This reduced delay is due to the lower gain (about 33 X per stage vice 1000 X for a single stage) of amplifier each stage allows the opamp to come out of saturation faster. Another way to get a slightly lower time out of saturation is to increase the supply voltage to the preamp.
Nice work!
bbsailorComment
-
bbsailor what would be the maximum supply voltage to the preamp that you would use ?
thanksComment
Comment