Electronics > Projects, Designs, and Technical Stuff
Help me understand and troubleshoot this early 70's circuit
Maxxarcade:
I am in the process of restoring a vintage electro-mechanical arcade game from 1973. It has a small sound board that makes two sounds for a crude simulation of crowd cheering. One sound is just filtered white noise (triggered by putting 18v B+ to pin 3), and the other sound is the same white noise but with an additional whistle added (triggered by pin 4).
The issue I am having is with the whistle sound. If I power the board at full voltage (30v AC) the whistle makes a short chirp noise when first triggered, and another chirp when the trigger input is released. It should make a continuous tone that starts shortly after the crowd noise, and fades out as the trigger is released.
If I either remove the D5 zener, lower the input voltage so that the zener is not regulating, or even just add a 470 ohm resistor between the zener/R26 junction and the rest of the circuit, it will work perfectly fine. It's like the B+ power rail has to be "soft" to allow for the oscillator to work. It also will not work if I use my bench power supply to input clean DC directly to the rail.
I'm pretty sure the problem is due to the original "X16A5332" transistors being replaced with 2N3904, but I really just want to understand why it behaves this way. Why would it need a soft/dirty B+ rail to operate correctly? And again, all functions are working correctly except for the whistle noise.
Note that there is an error in the schematic. Q4/Q5 emitters are missing their ground symbol.
The whistle sound is generated by Q3, Q6 and Q8. From what I can tell, Q6 and Q8 are a monostable multivibrator which is being triggered through Q3. C15 charges to a given amount depending on which trigger input is used, which in turn puts a positive voltage on the emitter of Q3, and also enables the white noise output from Q4 to pass to the audio amp. The higher the charge on C15, the longer the white noise stays active. VR2 adjusts the duration. The whistle will stop very shortly after B+ is removed from the trigger input, while the white noise will continue a bit longer.
A game with the same board can be seen/heard here, at around 43 seconds into the video (not my machine, just an example of how it should work):
Note that the game triggers the whistle 4 times. On the board itself, the whistle will go as long as the input is held to B+.
floobydust:
It's a clever circuit with many parts interacting to generate the sounds and envelope. It's not easy to figure out and might take a bit guessing.
Q8, Q6 look more like a one-shot ? (two capacitors for an oscillator and Q6 does not have one) but triggered by noisy Q3 over 7V on C15. Hmmm.
That would mean the whistle oscillator is Q1 but cleverly gated to use it as a (white noise) amplifier or oscillator, depending on the control voltage from Q2. As the voltage on C15 ramps up or down, it's also controlling noise envelope and oscillator frequency, so it's all complicated. I think Q2 on enables the oscillator by switching R8/D2 out of circuit. Then Q1 is just amplifying applause noise.
Hard to know what the gain of the original transistors was, if they were selected parts etc. 2N3904 has lowish beta at low collector currents and might not be ideal here. I would try a select high beta part for Q1. If the whistle oscillator is lazy to start up, due to low gain or the wrong control voltage, any ripple or disturbance on the power supply rail would help it start up. Maybe BC547B (flipped pinout). What is the zener voltage?
To make troubleshooting easier, you could connect a bench power supply to "pin 4" and vary the DC voltage up and down to see if the gates are in the right order and what the circuit does in slow motion at C15. In the video I hear a whistle up and down in freq. and applause having a VCF too.
Maxxarcade:
--- Quote from: floobydust on April 13, 2020, 02:29:36 am ---It's a clever circuit with many parts interacting to generate the sounds and envelope. It's not easy to figure out and might take a bit guessing.
Q8, Q6 look more like a one-shot ? (two capacitors for an oscillator and Q6 does not have one) but triggered by noisy Q3 over 7V on C15. Hmmm.
That would mean the whistle oscillator is Q1 but cleverly gated to use it as a (white noise) amplifier or oscillator, depending on the control voltage from Q2. As the voltage on C15 ramps up or down, it's also controlling noise envelope and oscillator frequency, so it's all complicated. I think Q2 on enables the oscillator by switching R8/D2 out of circuit. Then Q1 is just amplifying applause noise.
Hard to know what the gain of the original transistors was, if they were selected parts etc. 2N3904 has lowish beta at low collector currents and might not be ideal here. I would try a select high beta part for Q1. If the whistle oscillator is lazy to start up, due to low gain or the wrong control voltage, any ripple or disturbance on the power supply rail would help it start up. Maybe BC547B (flipped pinout). What is the zener voltage?
To make troubleshooting easier, you could connect a bench power supply to "pin 4" and vary the DC voltage up and down to see if the gates are in the right order and what the circuit does in slow motion at C15. In the video I hear a whistle up and down in freq. and applause having a VCF too.
--- End quote ---
The zener is 18 volts. I also tried some others from 13 to 22 volts, but the behavior is exactly the same. Any time the zener conducts and starts cleaning up the B+ supply, the whistle stops working.
Q6/Q8 do appear to be a one-shot, which I thought Monostable was just another name for?
Interesting take on Q1. I figured it was strictly an audio amp stage with an RC filter, but perhaps that could be part of the oscillator. Though if I remove Q3, the whistle stops. So that's definitely involved too.
Really the only "simple" part of this board is the white noise generator, which is a really common avalanche circuit. It's running all the time, and it seems like C15 controls how much of the signal comes out by varying the voltage at the anode of D3. I also need to decrease the value of R7 a bit to make the applause louder. These modern transistors don't put out as much noise as the originals likely did.
The applause doesn't change in frequency, just in volume. The more voltage on C15, the louder the applause. The whistle starts slightly after the applause and raises in pitch, then after the input is released it lowers in pitch and goes away, leaving the applause to fade out a few seconds later. When in the machine, the whistle gets triggered 3 or 4 times in rapid succession, which is why the pitch goes up and down.
I was doing some reading and heard about someone else using "C546B" transistors on their board, which are likely BC546B. Maybe I'll order some of those to try.
I'll try varying the voltage on the trigger input as well.
james_s:
That's neat, I love the old analog sound boards. I wish I had something more useful to contribute but I think you might be right about the modern transistors. The 2N3904 likely has far more gain than the original parts, it may be that something is saturating rather than oscillating.
Yes one-shot is another name for monostable multivibrator.
floobydust:
To make an analog bell bong sound, you take an oscillator with insufficient gain to keep running, it just rings once a pulse hits it, and then it decays and falls off to nothing.
So I think the crowd whistle is a weak oscillator that needs "bumps" from the noise-triggered one-shot, which is why pulling Q3 stops it. Q2 controls the gain of the Q1 stage.
I guess we're not sure if the newer transistors have too much or too little beta, assuming that is the problem. There are some high value resistors 10M, 2.2M etc. so I assumed the old parts had decent beta.
OP can check Q1-C voltage (at rest) and see if it's a few volts or too low due to the high beta. But only as few uA of bias in the first place.
Navigation
[0] Message Index
[#] Next page
Go to full version