Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Re: Ace Christy Mixer Question....
76 views
Skip to first unread message
kevi...@gmail.com
Mar 8, 2016, 6:20:27 PM3/8/16
to
Google isn't showing me a date for this thread. Sorry if it is a zombie, but maybe this info will help someone else:
Here is an online PDF of the manual for this device, which includes schematic.
http://www.e-radiocontrol.com.ar/downloads/vintage/servosyESC/Christy%20Mixer%20Users%20Manual.pdf
Yes, the mixer should work with any radio that uses standard servos (1.0-2.0ms pulses) If it is adjusted wrong (the pot on the oneshot) it might stop sending pulses to one servo and cause it to freeze.
A "how it works", and few tech notes on the schematic:
The op amp is an LM3900, which is a norton op amp. No other quad op amp will work in this circuit. A Norton op amp is unique in that it is easy to build an integrator with both inverting and non-inverting inputs, and this circuit depends on that. The 3900 is also an excellent choice for this low voltage single supply circuit. This Christy fellow, or whoever massaged the circuit into this form really knew his stuff.
The input buffers are NAND gates, not AND gates as drawn. So the central tap on the lower pair of buffers supplies an inverted signal B (B!) to the one-shot. Buffers are needed because the mixer relies on integrating 5V signals over time, and so any noise on the digital signals will corrupt the outputs. The mixer uses feedback from the outputs to compensate for variations in supply voltage. The RC filter at the top of the schematic helps clean up any noise on the supply.
Input A gets added into both outputs. Input B gets added to output X, and subtracted from the output Y.
The one-shot adds twice the 1.5ms neutral value to the inverted channel B. When B! goes low (start of servo pulse, B goes hi) the one shot begins a 3.0 ms output pulse on pin 10. But this is diode ANDed with B! so the actual pulse that goes to the lower integrator is 3.0 ms - T(B). So when B is 1.0 ms,the integrator gets a 1.5ms pulse, and when B is 2.0 ms, the integrator gets a 1ms pulse...so the one-shot provides a reverse rotation signal from channel B.
The one shot has a second function, it supplies a trigger reset voltage via the .01uF cap to the output comparators. More on this later.
So the upper integrator gets charged to a voltage representing A+B and the lower to A+ (B rotation REVERSED). This is from the positive current pulses supplied through the 100K resistors into non-inverting (current mirror on a Norton amp) inputs on pins 12 and 13.
When the one shot goes low, (3.0ms after the leading edge of input B) the two output comparators are triggered on by the negative pulse on their inverting inputs. So they both go hi. When they switch hi, they stop diverting the current from the 100K pots, (which should nominally be at 50K ohms) and it goes into the inverting input of the integrators causing them to integrate downward from the stored value created by the inputs. Since two inputs integrated upward, then the one output needs to integrate downward twice as fast to result in the same neutral. This means each input can contribute only half of full rotation of the servo, which is required to prevent over-driving the servo. This means if the 100K output scaling pots are adjusted wrong, the servo may see too long, or two short pulses.
Once the integrators return to near zero, they steal the hysteresis signal from the comparitors, causing the outputs to return low. Note the little "Gr" on the integrator output diodes. That means Germanium diodes are used here, because the Norton amp inputs are nominally one (Silicon) diode drop above ground, and the output can only go down to 100mV or so, so the integrator couldn't switch the comparator back low unless the diode there has a low drop. If you can't find a 1N34 to put in there, then a Schottky should work fine. Current is minimal so a mixer diode would work.
Because the circuit integrates upward and downward using the same supply voltage, the exact voltage value doesn't change the output timing. The one-shot, only shifts a percent or two with V+ between 4 and 6V...not too shabby.
Also note that if the servos use slightly non-standard timing, the only requirement will be to re-adjust the one shot to supply 2X neutral time pulses...the upward and downward integration action will scale the output timing to match the input timing. If the timing is much longer pulses than standard, the integrators may saturate, so larger input scaling resistors may be needed to reduce the output swing.
Finally, because the timing (which pulse comes first) between input A and B is unknown, the mixer may output bogus positions for the first pulse when the radio is switched on. Input A may not have arrived yet when input B switches the outputs hi, so it takes a cycle or two to sync up. This was not noticeable in the era when this mixer was used, because ALL the channels glitched at turn on with every radio I ever used.
Here is an online PDF of the manual for this device, which includes schematic.
http://www.e-radiocontrol.com.ar/downloads/vintage/servosyESC/Christy%20Mixer%20Users%20Manual.pdf
Yes, the mixer should work with any radio that uses standard servos (1.0-2.0ms pulses) If it is adjusted wrong (the pot on the oneshot) it might stop sending pulses to one servo and cause it to freeze.
A "how it works", and few tech notes on the schematic:
The op amp is an LM3900, which is a norton op amp. No other quad op amp will work in this circuit. A Norton op amp is unique in that it is easy to build an integrator with both inverting and non-inverting inputs, and this circuit depends on that. The 3900 is also an excellent choice for this low voltage single supply circuit. This Christy fellow, or whoever massaged the circuit into this form really knew his stuff.
The input buffers are NAND gates, not AND gates as drawn. So the central tap on the lower pair of buffers supplies an inverted signal B (B!) to the one-shot. Buffers are needed because the mixer relies on integrating 5V signals over time, and so any noise on the digital signals will corrupt the outputs. The mixer uses feedback from the outputs to compensate for variations in supply voltage. The RC filter at the top of the schematic helps clean up any noise on the supply.
Input A gets added into both outputs. Input B gets added to output X, and subtracted from the output Y.
The one-shot adds twice the 1.5ms neutral value to the inverted channel B. When B! goes low (start of servo pulse, B goes hi) the one shot begins a 3.0 ms output pulse on pin 10. But this is diode ANDed with B! so the actual pulse that goes to the lower integrator is 3.0 ms - T(B). So when B is 1.0 ms,the integrator gets a 1.5ms pulse, and when B is 2.0 ms, the integrator gets a 1ms pulse...so the one-shot provides a reverse rotation signal from channel B.
The one shot has a second function, it supplies a trigger reset voltage via the .01uF cap to the output comparators. More on this later.
So the upper integrator gets charged to a voltage representing A+B and the lower to A+ (B rotation REVERSED). This is from the positive current pulses supplied through the 100K resistors into non-inverting (current mirror on a Norton amp) inputs on pins 12 and 13.
When the one shot goes low, (3.0ms after the leading edge of input B) the two output comparators are triggered on by the negative pulse on their inverting inputs. So they both go hi. When they switch hi, they stop diverting the current from the 100K pots, (which should nominally be at 50K ohms) and it goes into the inverting input of the integrators causing them to integrate downward from the stored value created by the inputs. Since two inputs integrated upward, then the one output needs to integrate downward twice as fast to result in the same neutral. This means each input can contribute only half of full rotation of the servo, which is required to prevent over-driving the servo. This means if the 100K output scaling pots are adjusted wrong, the servo may see too long, or two short pulses.
Once the integrators return to near zero, they steal the hysteresis signal from the comparitors, causing the outputs to return low. Note the little "Gr" on the integrator output diodes. That means Germanium diodes are used here, because the Norton amp inputs are nominally one (Silicon) diode drop above ground, and the output can only go down to 100mV or so, so the integrator couldn't switch the comparator back low unless the diode there has a low drop. If you can't find a 1N34 to put in there, then a Schottky should work fine. Current is minimal so a mixer diode would work.
Because the circuit integrates upward and downward using the same supply voltage, the exact voltage value doesn't change the output timing. The one-shot, only shifts a percent or two with V+ between 4 and 6V...not too shabby.
Also note that if the servos use slightly non-standard timing, the only requirement will be to re-adjust the one shot to supply 2X neutral time pulses...the upward and downward integration action will scale the output timing to match the input timing. If the timing is much longer pulses than standard, the integrators may saturate, so larger input scaling resistors may be needed to reduce the output swing.
Finally, because the timing (which pulse comes first) between input A and B is unknown, the mixer may output bogus positions for the first pulse when the radio is switched on. Input A may not have arrived yet when input B switches the outputs hi, so it takes a cycle or two to sync up. This was not noticeable in the era when this mixer was used, because ALL the channels glitched at turn on with every radio I ever used.
0 new messages