+3 Audio hates me with a passion
Hihi
Since nobody seemed to have a spare +3B board, I've been trying to replace the audio entirely, aiming for a speaker inside the machine
Since pins 1, 4 and 5 on the AY are connected together, I just cut the track leading from pin 1 and connected a small LM386 based amplifier direct to chip
This one in fact: http://web.mit.edu/6.s28/www/schematics/lm386.htm
However, it didn't work, if I feed it the output from the tape socket I get (very loud) 48K audio but nothing at all from the AY, it just autoadjusts to loud background noise (which it also does with nothing at all connected)
It's possible from what I've read the AY output just isn't pwerful enough but I thought I'd hear something even if very quiet and I don't..
Some people have suggested lifting the pins instead of cutting the track, would that make any difference? from what I can see on the other side they're just connected together on the bottom and don't seem to go anywhere..
Suggestions would be helpful...
Thanks
(PS: I tried the single resistor fix for TR5 but it was still badly distorted so I took that out again)
(PPS: Yes, I also posted this to comp.sys.sinclair but nobody answered so here it is here -.o)
Since nobody seemed to have a spare +3B board, I've been trying to replace the audio entirely, aiming for a speaker inside the machine
Since pins 1, 4 and 5 on the AY are connected together, I just cut the track leading from pin 1 and connected a small LM386 based amplifier direct to chip
This one in fact: http://web.mit.edu/6.s28/www/schematics/lm386.htm
However, it didn't work, if I feed it the output from the tape socket I get (very loud) 48K audio but nothing at all from the AY, it just autoadjusts to loud background noise (which it also does with nothing at all connected)
It's possible from what I've read the AY output just isn't pwerful enough but I thought I'd hear something even if very quiet and I don't..
Some people have suggested lifting the pins instead of cutting the track, would that make any difference? from what I can see on the other side they're just connected together on the bottom and don't seem to go anywhere..
Suggestions would be helpful...
Thanks
(PS: I tried the single resistor fix for TR5 but it was still badly distorted so I took that out again)
(PPS: Yes, I also posted this to comp.sys.sinclair but nobody answered so here it is here -.o)
Post edited by Hideki on
Comments
It should be but on the original +3 it isn't -.-
I'm guessing some sort of preamp is required but really I'm not sure and I'm not exactly being flooded with useful answers here or on usenet, lol
Not an expert with micro computer electronics but i'm sure the Gate Array/Pin 48 and 52 won't get anything from the AY after just cutting the track from pin 1 on the AY. I'm guessing that a gate array does switching (anyone?), and important.
From experience in building low power audio pre-amp circuits, the best start to fixing distorted audio would be to test the voltages of TR5. Measure the Collector, Base and Emitter leads of TR5. Its ideal to have half the supply voltage on the collector. If its mis-biased, you get more voltage swing on the + than the - (or vice versa), which causes clipping from no headroom (balanced voltage swing).
You could also try removing TR5 and replace it with a socket. Then try dropping in some replacements until you find one that is biased better. I'd start with BC548B's or 2N3904's as alternatives. I wouldn't expect much improvement without biasing though unless you get lucky. One more trick would be to plug in TR5 backwards and see what happens, with collector and emmiter reversed, this may lower the gain somewhat.
If I had some time, I'd replace TR5 with a J201 FET and spend time biasing it with trim pots. FET's have much better signal/noise ratios and lots more headroom.
Frank
My ZX Art Music Page
Carlos Michelis Theme
http://www.muzique.com/lab/lowvolt.htm
http://www.muzique.com/lab/reverse.htm
And I guess you've already visited here regarding the +3 sound.
http://www.secarica.ro/html/plus3_hardware.html
My ZX Art Music Page
Carlos Michelis Theme
Ah yes, thanks ^^
I did try the 220R on TR5 fix mentioned there (I think)
I'm going to try the resistor fix mentioned in another thread tomorrow (or the two resistors fix), people seem to have differing ideas about what values to use mind
Replace R72 with a 1k.
Lift the emitter leg of TR5 (ground connection), onto the component side.
Solder a 220 and 10k resistor together and attach TR5's emitter leg to free end of 220 resistor.
Solder 10k/220 resistor join into where you removed TR5 emitter.
Solder free end of 10k resistor to Base of TR5 (middle pin i'm sure, but google the datasheet for BC239B to make sure).
If after this there is still distortion, temporarily replace R62 with a 10k potentiometer and tweak until you measure half the supply voltage (5v, but test it first) between the collecter pin and ground of TR5. Then disconnect the pot, measure its resistance and replace R62 with a resistor to the nearest value you measured.
Edit: Here's a drawing, the thin outline is the +3 pcb holes for TR5, the thick outline is the actual TR5 component. NOTE: The orientation of the +3 pcb may be different, I haven't seen a picture of TR5 from the component side.
My ZX Art Music Page
Carlos Michelis Theme
Just soldered up a little board which I'll try shortly
[edit edit]
Incidentally I've noticed that if you change R62 too much you get a lot of noise on the picture when sound is playing (which is unfortunate as I discovered soldering a 680R resistor across that one in parallel and a 68R across R72 as someone else suggested you get perfect sound at the normal volume, but with noise on the picture -.-)
[edit edit edit]
Okay, added little board and replaced resistor with 1K, still somewhat distorted, need to hunt down a 10K pot now to twiddle with R62, lol
Fixed my diagram (TR5 reversed).
Can you measure the voltages of TR5's collector, emitter and base, and post them here.
Edit:
And measure the the leg of R62 which is furthest from TR5 collector (before it goes through R62 to TR5 collector)
Thanks
My ZX Art Music Page
Carlos Michelis Theme
I'll check those shortly although there's a 10K pot in there now so I'm not sure how much good it'd do measuring it -.-
In either case, the audio is still distorted whatever I set it to, it's better than it was but still not as good as the +2 audio (or fuse come to that)
I'm not sure I'm getting meaningful readings from TR5 unless two of them should be less than 1v and the other should be a touch over 3, if they should be like that, R62 is 5v on one side, the other varies depending what I set it to, lol, do I need to set that to 2.2 and take the measurements then? if so I'll do that
Set the pot to roughly 2.5v on the collector of TR5. Once you've got 2.5v, disconnect the pot (without moving it), and measure its resistance. Then you can replace the pot with a single resistor near to the value of the pot.
Voltages of TR5 should be from 0 to 2.5v on the various legs. Its common to read 0.4v etc.
Then you can use that 10k pot to temporarily replace other resistors to tweak, I'd probably tweak the 10k attached to the 220R as well.
I've got my +3 out of the loft, and am going to test the voltages of TR5, before I've tampered with it. I was reluctant to mess with my +3A, as its brand new never used, still unopened in its original packaging/box.
My ZX Art Music Page
Carlos Michelis Theme
Unfortunately, I'm unable to plug my +3 into a tv yet, I need to borrow a CRT tv. I only have an LCD at the moment.
More information on the type of bias used in the original +3A.
Collector-to-base bias.
If your not sure about the equations, i'm sure someone better at maths than me can help. But Vbe means the base-emitter Voltage, and Rb is the resistor connected to the Base of the transistor etc.
The changes I suggested have given us a collector to base bias and voltage divider bias.
Edit again:
Remove the new 10k attached to the new 220R, and try a 100R - 470R in place of R72.
The R204 10k and R200 820R of the +3B in parallel equals 757.85 ohms, its 2.2k in the +3A.
Reports of 75 and 220R resistors in parallel with 2.2k R72 equals 73-200 ohms for replacing R72.
My ZX Art Music Page
Carlos Michelis Theme
I reversed the polarised caps in my schematic, compared to the schematic i found HERE
There's a couple of small differences between Sinclair 128 and Amstrad +2.
22nF capacitor between pins 17,3,19(+5v) and 6(gnd) of AY chip on Sinclair.
And small changes for inbuilt cassette from Pin 34(Ear) of ULA.
From Pin 35(Mic) of ULA going to MC1376 pin 5, 68k resistor on +2 instead of 6.8k Sinclair 128.
MIC socket is audio out socket on +2.
Polarised caps are reversed, (could be mistake in schematics).
FWIW, I had both a Sinclair 128 first and then a +2, the Sinclair 128 had the best sound and very reliable tape loading through the Ear socket.
My ZX Art Music Page
Carlos Michelis Theme
My ZX Art Music Page
Carlos Michelis Theme
My ZX Art Music Page
Carlos Michelis Theme
Components list
C1 100n
C2 100n
C3 100n
C4 10n
C5 1uF
C6 1uF
D1 1N4148
R1 1.8k
R2 1.8k
R3 1.8k
R4 560R
R5 330R
R6 180R
R7 10k
R8 10k
R9 680R
R10 1k
R11 18k
R12 6.8k
TR1 BC239
My ZX Art Music Page
Carlos Michelis Theme
Have you actually made this?
I haven't had time to build it yet, but it should, in theory replace the +3a sound entirely with the circuit from the Sincalir 128. Providing the correct components are removed and tracks cut on the +3a board.
So I cannot verify 100% yet, how well it will work.
My best suggestion for you with all that hardware in there, is to keep lowering R72 until it sounds clear. And adjust R62 if theres noise in the picture.
My ZX Art Music Page
Carlos Michelis Theme
http://www.youtube.com/watch?v=nl_mg9hpoEw
Although it may be done with only two resistors (not tested):
- Lift a pin, or remove, R42 (location available in the video)
- Cut the track as showed in the video.
- Solder the 1K resistor as in the video.
- Solder a 27K resistor between Base and GND.
schombi.de/my-retro-collection.html"
It didn't seem to work at the time, the output was heavily low pass filtered and still distorted. So I put it aside until revisiting it last night.
The mod I tried was to add a 12K resistor between the base of transistor TR5 and ground, and a 330 ohm resistor from the emitter of TR5 to ground. This was intended to modify the +3 circuit to be more like the +3B circuit. This has been suggested elsewhere in this thread, but for the sake of clarity I'll explain it here.
The +3B circuit on www.secarica.ro has a 39K resistor from the base of TR200 to the collector, and a 10K resistor from the base to ground. These form a bias network. To achieve the same potential divider on the +3, which has a 47K base to collector resistor, the nearest value is 12K.
The +3B circuit has a 1.5K resistor from the collector to +5V and a 220 ohm resistor from the emitter to ground. To achieve the same gain on the +3, which has a 2.2K collector resistor, the nearest value for the emitter resistor is 330 ohm.
With these changes, the output was still slightly distorted and heavily low pass filtered. The component causing the low pass filtering on the +3 is C37, a 100nF capacitor connected between the collector of TR5 (which is the output of the audio amplifier) and ground. This capacitor forms a low pass filter with R62, the 2.2K collector resistor. The low pass filter has a cutoff frequency of 723Hz. Frequencies above this value will be attenuated at 6dB/octave.
So last night I tried changing C37 for a smaller capacitor. The best standard value capacitor to let most of the audio range through was 4.7nF. Together with R62 this forms a low pass filter having a cutoff frequency of 15.392kHz.
With this change in place, the audio output now has the high frequencies, but there is still a little distortion. It's not unpleasant sounding, in fact it's quite nice, but it's still not right.
I haven't got a +3B to check, but there is a discrepancy between the circuit on the www.secarica.ro pdf and Amstrad's +3B service manual. The service manual shows the 39K resistor connected between +5V and the base of the transistor, and the 10K resistor connected between the base of the transistor and ground. In other words a conventional bias network or potential divider driven by the power supply. Whereas the pdf shows the 39K resistor connected between the base of the transistor and the collector of the transistor. This may have been because the +3 circuit used collector to base biasing, and the author of the pdf copied this across to the +3B circuit diagram and added the extra components without noticing the 39K resistor is actually connected to +5V. Or it may be because actual +3B computers are wired this way round rather than as shown in the service manual.
Assuming the service manual is correct, this would hold the base of the transistor at around 1.02V, which may be enough to prevent the distortion I am hearing with the modified collector to base circuit currently running on my +3. When I am tired of hearing the slightly distorted sound I have currently, I will try moving the 47K resistor to +5V instead of the collector of TR5.
But the service manual and pdf document do agree on the value of C37 in the +3B (100nF) which in conjunction with the 1.5K collector resistor form a low pass filter of 1.061kHz. This circuit will still sound noticeably filtered, and I would suggest replacing C37 on the +3B with 4.7nF to move the cutoff frequency to 22.575kHz.
Edit: (8/4/2013) I moved the 47K resistor to +5V very late last night. Circuit is working fine now.
For the differences between the +2A/+3 sound circuit and the +2B/+3B circuit see my fixing the +3 sound page. There are six resistors which were changed and two resistors added.
As far as I'm aware there is currently no actual physical documentary evidence of the +3B ever having actually existed.
Before I get yelled at, I know two people who claim to have seem one back in the day but still, I'm waiting for photographic proof :)
Thanks for the info, guesser. Nice page! I've wanted a +3 with sound as good as my Spectrum 128 ever since I got my +3 some time in 1992. I never thought it would take until now to achieve it! Comparing the two mods, they achieve essentially the same thing using slightly different components. The potential divider for the bias network in the +2B circuit is 39K:10K, whereas in the circuit I used it is 47K:12K, using the original 47K base to collector resistor with the collector terminal moved to +5V, and a new 12K resistor to ground. The collector and emitter resistors in the +2B circuit are 1.5K and 220 ohm, whereas in the circuit I used they are 2.2K and 330 ohm, using the original 2.2K collector resistor and a new 330 ohm emitter resistor. These produce broadly the same bias point and gain structure as the +2B circuit. I was just trying to scale up the new components so that the existing resistors could be reused. I didn't replace any of the other +3 resistors.
That takes care of the distortion, but to finish off the job you need to remove the low pass filtering still present in the +2B circuit. C37 should be changed from 100nF to 4.7nF.
http://www.amibay.com/showpost.php?p=380700&postcount=13
So the circuit sounds about the same as a +2B output would do, possibly a little worse, possibly a little better. The smaller 820 ohm pulldown resistor R200 in the +2B may make the problem of interaction between the three channels a little better, but I don't think it would go away completely. I might try replacing R72 with 820 ohm but it seems the way to get the same sound quality as the 128 / +2 is to put series resistance on all three channel outputs and that means modifying the +3 circuit board extensively since the three channel pins are connected together with PCB traces on the +3.
Correct, it's crap design and I'm fairly sure this is the main reason why emulated AY doesn't sound like a real +2B.
I don't have a spare AY chip to do experiments on though and I don't want to rip my +3 apart to do so :)
Well I'm not really sure where to go from here. I've got a +2 on which I've done a 3 channel AY output mod, putting the three channels on different pins of a 5 pin DIN connector. On that machine it's easy to get at the three channels because there are series 1K8 resistors connected to each channel, I just desoldered them, soldered wires to a piece of Veroboard and put small buffer circuits on that.
The PCB traces joining the three channels together on the +3 are on the solder side of the board, so cutting them can be done without desoldering the AY chip. From there it's a matter of attaching wires or resistors directly to the pins. But to get the sound of the 128 / +2 circuit you'd have to change the rest of the circuit extensively. It would be almost as easy to do the 3 channel mod as to copy the +2 circuit.