A Bit of Information about the Crumar Bit99 (AKA Unique DBK)
Debugging a unit with no sound:
Recently I was given a Crumar Bit99, in the guise of a Unique DBK. The display changed when I selected different patches, but there was no sound when I pressed keys. A web search turned up several related schematics: Crumar Bit One and also Crumar Bit 01, which seems to be closer. Neither of these is an exact match for the Bit99, but the circuits are very similar, and those schematics are a lot better than none at all. I looked at the DAC output, and the analog signals feeding into the output board, but none of the signals changed at all when I pressed keys.
The Bit99 uses an 8031 (a romless 8051) plus external EPROM and RAM for the main control CPU. It also has an 8051 mounted on the underside of the keybed, for keyboard scanning and velocity measurement. There are two cables connecting the keyboard to the rest of the unit. One has just four wires and connects to the power supply board. This one is called JO. I think the pinout is:
Left pin - 5V
Other three pins - Ground
The other cable has six wires, and connects to the front panel pcb, at a location called JN. I think the pinout of this one is:
1) 59 KHz clock. I think this is being fed to the keyboard.
2) Floating? seems to have a triangle wave on it, maybe just noise from the next signal over?
4) Seems to be Ground
5) Keyboard serial data stream. Std async serial data (lsb first), two bytes, sent only when a key is pressed, or released. Baud rate appears to be 31.25 KHz (the MIDI baud rate). Looks like first byte is the key number, starting at 18h (24 dec) with the lowest key, and incrementing from there. The second byte varies, so it's probably related to velocity on key press. On key release, the second byte is 00h, which makes sense.
6) 2.0 MHz clock. Not sure where this comes from.
In my unit, none of the signals on this connector changed when I pressed keys, so I figured there must be a problem with the keyboard itself. Turns out one of the traces from the 8051 to its 12 MHz crystal was eaten away, so there was no clock to the keyboard scan MCU. A few other traces also looked a little "iffy" so I fixed all of them by adding wires to jumper the marginal areas. I re-assembled the keyboard and the unit started to make noise. I downloaded the first ten patches using MIDI and my PC and started playing chords, and selecting different programs. Anyway, all six voices are working, and I am now up to 40 programs downloaded. I guess the 8031 wasn't fast enough to keep up with downloading the entire set at once.
Some info on setting the potentiometers:
There are many trimpots in the Crumar Bit 99, and so far I have found nothing on the web about how to adjust them. If your unit is working OK, I would suggest not adjusting anything, but in some cases you may have no choice. Also, there is no way you will be able to adjust the F1 and F2 trimpots without an oscilloscope. After fooling with the pots a little in a Bit One, which is pretty similar, we have learned a little about what they do. I have not found a schematic for the Bit 99 on the web, but it is the same as a Bit 01 (Not the Bit One) which was a rack mount version. Here is some information on the trimpots:
P115 PC Bd Pots:
Trimmer 5 (4.7K above IC4) I think this would adjust the high pulse width of the F1 clock. Should be on the order of 250 nSec.
Trimmers 7 and 8 (above IC7) Adjust the high and low widths of the F2 clock. Should be on the order of 250 nSec each, although the low width can be as short as 150 nSec and still meet the 8253 clock requirement. The high width needs to be a minimum of 230 nSec. Note that the width changes with modulation, such as moving the pitch wheel. The Bit One schematic shows a frequency of 2.0 MHz for both F1 and F2. This seems to be the nominal value with no modulation and the pitch wheel centered.
Trimmers 3,4 and the one next to IC6 seem like they would adjust the bend amount. The Bit One circuit is a little different. We learned that it's important to have the bend amount match for F1 and F2, so that the clocks stay in sync as their frequency is raised or lowered. Otherwise a chord with notes using both oscillators will be out of tune as the pitch is raised or lowered.
Trimmer 2 (Next to DAC) This is the DAC scale/range pot. Probably there is some magic combination of switch presses that invokes a DAC range set mode. Without that, it's best to leave this one alone. We could guess that the DAC range should be 5V, and try to set some parameters to maximum, like the VCA or filter envelope amount, then set the pot to get a 5V value from the DAC at the peak of the envelope, but we would just be guessing.
+Vt adjust trimpot: (Next to BC173C transistor and 1K resistor) This pot should be adjusted with an oscilloscope to get the best triangle wave shape. Each oscillator generates a sawtooth, a triangle, and a pulse waveform. You can see these at the inputs to the 4016 IC's on the DCO board. Find the triangle one and monitor it with an oscilloscope, then adjust the trimpot for the best waveform. You will quickly see what the pot does. This is an easy setting if you have a scope. Without a scope it would be pretty hard to get it right, although you might be able to do it by ear. I don't think I would trust myself to do it that way.
Voice Board Pots:
Trimpots 2,1,6,5,10,9: (One for each of the six voices) These control the gain/scale of the CEM3328 VCF cutoff frequency. Possibly they could be set using the keyboard tracking mode of the filter, to get proper tracking.
Trimpots 13,14,15,16,17,18: (One for each of the six voices) These control how much voltage is fed to the filter resonance control input pin on the CEM3328. We were able to set these by ear, although I am sure that's not how the factory did it.
Trimpots: 3,4,7,8,11,12: (One for each of the six voices) This seems to be some kind of VCA offset adjust. We monitored the voltage at pins 12 and 5, which are separate outputs on the Bit One, and adjusted the pots with a low level signal coming in, to get symmetric swing on the unbuffered outputs at 12 and 5, above and below 0V. Adjusting the pot moves the output DC level up or down relative to ground. On the Bit99, since three of these pins are tied together, it will be a little more difficult. If possible, arrange to have a signal coming in on just one of the three channels, and adjust the trimpot for just that channel so that the output swings evenly above and below 0V, then feed in a signal on a different channel and adjust that trimpot for no offset. I have not tried this so it may not work. The signal coming in should be at a low level to make it easier to see when the offset is low.
I recently tried to change the code to implement LFO modulation of PW instead of velocity. Actually, I was working with a BIT01, not a BIT99, but they are quite similar. I ended up giving up on that project, as the steps were too coarse, due to the 8-bit DAC and 8-bit LFO value. But I did spend some time figuring out some things about the code, which I have decided to share here. Here is an archive containing the BIT99 code, after disassembling it and spending some time to figure out some of how it works. And here is an archive containing the BIT01 code. It also contains a binary image for the C2 EPROM which has been changed to not come up in OMNI mode after reset. The .asm files in these archives are valid input for TASM32, and will produce binary images that are identical to the EPROMs which I started from. (except the place I changed to disable OMNI mode) The Bit99 code has a lot more comments in it. Many of those comments would apply to the BIT01 code as well, as much of the code is either similar or the same.
BIT99 schematics are available
I was able to purchase a pretty nice scan of the actual BIT99 schematics, not BIT one or BIT 01, which are a little different. The seller has an ebay ad which he states will be running permanently. Here is a link to his ebay ad, which has fixed price. I have no connection with this person, and no interest in any sales.
For some reason, when I click on this link, the last "1" is dropped off, making the item number invalid. But if I paste the text into my browser, the whole number is used. Go figure.
If that link is no longer valid, you can contact the seller at this address:
Copyright © 2007-2016 by TaunTek.com. All rights reserved.