MAMEWorld Forums - EmuChat - post Midway Gun Fight netlist audio discussion

Always good to see some fine tuning when necessary for the games that use analog audio hardware for generating the audio sound effects. It is interesting that maybe some of the pcbs had been repaired using different components which could be why some sound effects have a different pitch frequency etc. That was one of the things that Gyrovision (Chris L.) had mentioned to me when he was working on his Congo Bongo pcb and had noticed that previous/prior pcb work from a pcb owner had used wrong parts when repairing Congo Bongo pcb. Gyrovision repaired the Congo Bongo pcb to parts shown on Congo Bongo service manual. After the repair work was completed, the game sounded the way he recalled it should sound since this was a game he liked himself. He then made audio recordings for external audio samples and sent to MAMEdev. This was many years earlier between 2006 to 2009 iirc.

*from Bannisters*
--
Colin Howell: ok, some weirdness...
Colin Howell: I was looking over Gun Fight again, after doing the 280-ZZZAP work, to see if I might want to do any tweaks (for example, a proper emulation of its noise generator using netlist's new Zener and noise modules)
Colin Howell: and I checked over the transistor models again (Vas had remarked that I didn't pick a specific BJT type, and I'd replied that the actual type used was kind of obscure)
Colin Howell: So I looked at that transistor again, the Amperex A-138, and the small amount of info I have on it.
Colin Howell: It seems like the BC548C would be an OK substitute; from the parameters for the A-138 I do have, the BC548C comes pretty close.
Colin Howell: Here's the thing...
Colin Howell: The default NPN I was using has a current gain of 100 (collector current is 100 x base current).
Colin Howell: The BC548C model has a gain of 466.
Colin Howell: For actual transistors, the gain is a range; it varies from one device to another. For the A-138, max gain of 650 is listed. Min is not, but the next lower grade (A-137) has a max of 415, so A-138's min should be around 400.
Colin Howell: Turns out, if you use transistors of gain 400-500 in the Gun Fight circuits, it doesn't just boost sound effect volume, as you might guess. It significantly changes the shot sounds.
Colin Howell: Because when a shot sound starts, its amplifying transistor first *saturates* for the initial attack, when the switch is on. You get a pulse, then silence as the transistor output is forced low.
Colin Howell: Then, when the switch goes off and the current drops a bit, the transistor leaves saturation and returns to the linear forward-active region...
Colin Howell: ... and then it starts amplifying normally, so now you hear amplified noise.
R. Belmont: interesting, and it makes sense once you explain it.
Colin Howell: The effect is an initial crack, a momentary pause, and then a noise burst that decays.
Colin Howell: For a while I was thinking I'd found some sort of modeling bug, but I got the same effect simulating the circuit in LTspice.
Colin Howell: But this only happens when you use high-gain transistors (say, 400-500). Low-gain ones (say, 100-200) won't do this.
R. Belmont: It's good that LTspice confirmed our behavior.
Colin Howell: Which makes me wonder if some people who've repaired their Gun Fight sound boards and replaced the hard-to-find A-138s might have changed their shot sounds without knowing it.
Colin Howell: Because when I check videos, I think I hear this effect in some but not in others.
Colin Howell: (it seems pronounced in my simulation but hard to catch in the videos.)
R. Belmont: Colin: part of that is that most videos are micing a cabinet, which can do weird things to the frequency response.
Colin Howell: yup. I'm thinking of asking on the arcade-museum forums to see if anyone there with a Gun Fight can comment.
Colin Howell: also, since it seems BJTs can lose gain over time, could Gun Fights with their original transistors lose this effect as they age...
R. Belmont: Absolutely. Plus old capacitors get leakier over time and can short, and at least some kinds of resistors will drift (usually upwards)
-