• Please review our updated Terms and Rules here

Stand-alone TU55 controller?

MattisLind

Veteran Member
Joined
Sep 30, 2013
Messages
1,158
Location
Stockholm, Sweden
Now that Anders and I have got this big pile of DECtapes (or LINCtapes) each it would be very useful to recover any software from these. Getting the PDP-9 to run is a major task. And the PDP-12 is a five year plan.

I thought it would be useful to have a stand alone TU55 controller that one could connect to a PC via USB. And it would be a new interesting project now that my PC04 reader project is more or less finalized (it works fine with a PC but I like to test it with a 11/03 and a M8028 card and load some actual software).

From my understanding the TU55 is a fairly simple device. SELECT lines, FWD, STOP, REWIND, GO and a few other digital signals. All these signals on one single board. Then the actual R/W head data on a dual board. The analog R/W head data is feed into either a G882 or G888 module. The G882 seems to be just a discrete non-adjustable version of the G888. Not being an analog expert, but I think it should be possible to replace the old op-amps / discrete transistors with some modern simple op-amp and also a circuit that prolongs the pulse width.

As far as see it the period of the timing track is quite low and would probably be possible to handle with an AtMega chip.

So what do you think? Has somebody already done this? Am I missing some severe problems related to the TU55 and the format of the tape? Any input is appreciated!
 
Hi All;
MattisLind, I Think that put in the Right/correct designer's hands, yes, it could be done..
"" Getting the PDP-9 to run is a major task. And the PDP-12 is a five year plan."" And making one of these is not a Major Task..
"" The G882 seems to be just a discrete non-adjustable version of the G888. "" So the G888 is an adjustable and possibly a later unit/design..
"" From my understanding the TU55 is a fairly simple device. "" I would Never call something simple, unless You were the origional designer..
"" be possible to handle with an AtMega chip. "" Maybe make it so a wider variety , different brands could be used..
If, this could be done, Could it Read TU56 and TU58 Tapes, which would make it be usable for a more favorable market..
My thoughts..

THANK YOU Marty
 
Has somebody already done this?

I went through a fairly long design brainstorm early this year. A MC3480 floppy disk data separator chip would be adequate
since you don't need compatibility with the quirks of the G882/G888.

Years ago, I built a simple interface with a FIFO to a parallel port. DECtape is self-clocking, and it isn't difficult to decode the
tracks in software.
 
While you guys are in brainstorming mode and before any hardware ideas start to gel, how about a device that could also read those tapes that due to damage or age cannot be read by the original hardware.
 
Hi All;

"" I went through a fairly long design brainstorm early this year. ""
Al, and You didn't share it with the Rest of Us..
Another one of Your Pearls of Wisdom !!
THANK YOU Marty
 
I went through a fairly long design brainstorm early this year. A MC3480 floppy disk data separator chip would be adequate
since you don't need compatibility with the quirks of the G882/G888.

I cannot find the MC3480 floppy disk separator. It is listed as a memory controller in the sources I can find. Are you sure about that part number? Intersil listed the HD-6409 and HD-15530 as Manchester encoder / decoder chips. But they are expensive and very expensive...

This is my thinking, correct me if I am wrong:
If we assume that we have the clock signal (we disregard the case where the clock signal is damaged) and then the data signals from the data heads are limited and produce a logic pulse signal for the positive going transition from the head and then a another logic pulse signal for the negative going transition (reflecting the switch of magnetisation on the tape). Feeding those into a RS flipflop would then recover the signal. We use the clock to sample that signal and then I guess that we have the data.

Years ago, I built a simple interface with a FIFO to a parallel port. DECtape is self-clocking, and it isn't difficult to decode the
tracks in software.

Yes, manchester/PE coding is self-clocking so a brute force variant is to sample all five channels at some nice rate and then post process it to find the framing. The advantage would be that it should be possible to recover even a damaged tape. But I think I rather have it all processed in the micro controller. Ideally it would present it self as a USB connected mass storage device if that would be possible.
 
If, this could be done, Could it Read TU56 and TU58 Tapes, which would make it be usable for a more favorable market..
THANK YOU Marty
TU56 is very similar to TU55. They as far I see use the same interface. The TU58, aka DECTape II is something completely different. They use a DC100 style cartridge. The existing controller speaks serial port very well so I don't think there is any use for another controller.
 
This has been progressing very slowly. First I needed to order new AC capacitors to replace the two PCB filled ones.
XFBE6u1.png


I don't think that Clorinol would be such a good trademark today... The new one is substantially smaller but had almost the same thickness and would mount easily in the capacitor compartment of the drive.

Then on to testing the drive. It worked sort of. In local mode the left motor was making the correct speed and torque, but the right didn't wind at nearly the same speed and when stopping winding it wasn't able to give the right amount of torque to have the tape tensioned so that the tape nearly jumped off.
The problem? It turned out the one of the G850 modules were bad. One diode in the rectifier bridge was open circuit. Fixing that and the TU55 winded nicely in both directions.

To analyze the signal from the tape I built a really horrifying mockup based on a very old abandoned project. Mostly an op amp and a +/- 5V regulator.
Cfr7Gxo.jpg


To be able to test the reed relays of the G851 module need be turned on. This can be accomplished by grounding the signal at A09:R. This won't cause any harm.

With this in place I was able to get this on my scope:

vREjODY.png


The upper trace is the timing track and the lower is the mark track. The period of the timing signal is 35 us which seems about right. Looking into this a while I got that if the mark track signal is sampled at the time when negative slope of the timing track signal crosses zero we can get either the peak or bottom of that mark signal. If we do that we get a sequence something like 0, 0, 1, 1, 1, 0, 0, 0, 1, 1, 1, 0, 0, 0.

The good thing is that this is probably correct since the data mark code is octal 70 or 111000 in binary which is repeated on the mark track when there data in the data tracks.
Next step would be to amplify the signal some more and then digitize it. Use the negative going edge of the timing signal to sample the data and mark tracks and have a small micro controller process the resulting data.
 
Use the negative going edge of the timing signal to sample the data

Look at the theory of operation of magnetic tape read channels first. What you are describing is a simple amplitude discriminator
which will have very poor performance in the presence of tape dropouts and amplitude variations which will be a big problem on
old tapes.
Read channels normally differentiate the incoming signal so that what you are looking at is the rate of change of the input signal
instead of the amplitude.
DECtape has an interesting problem for phase-encoded data recovery in that the data clock rate varies, so your sampling window varies too.
 
Look at the theory of operation of magnetic tape read channels first. What you are describing is a simple amplitude discriminator
which will have very poor performance in the presence of tape dropouts and amplitude variations which will be a big problem on
old tapes.
I mostly looked at the G882 and G888 cards. But I think I see what you mean. If I would use a differentiating circuit and a peak detector on the timing track I would have short pulse at the where the timing track signal crosses zero. That pulse would then be less dependent on the actual amplitude. If I then separate the negative and positive going pulses and feed them into a RS flip-flop I would reconstruct the signal.

Do you have any good reference to literature on tape read channels?

However I did digitize the signal using the simple approach and then feed it into my logic analyzer. The output looked fine and mostly consisted of sequences of 111000111000111000 which is the DATA MARK. Then there were a sequence which seems to be the block header:

Code:
1 1 1 0 0 0 1 1 1 0 1 1 1 1 1 0

1 1 1 1 1 0 1 1 1 1 1 0 1 1 1 0 

1 0 0 1 1 0 0 1 0 1 0 1 0 1 0 1

0 1 0 1 0 1 0 1 0 1 1 0 0 1 1 0

1 0 0 0 1 0 0 0 0 0 1 0 0 0 0 0 

1 0 0 0 0 0 1 0 0 0 1 1 1 0 0 0

If group that into six bit words and look up the different mark codes:
Code:
111000 DATA MARK
111011 PRE-FINAL MARK
111011 FINAL MARK
111011 PCC MARK
111011 REVERSE LOCK MARK
101001 GUARD MARK
100101 REVERSE BLOCK MARK
010101 INTERBLOCK SYNC MARK
010101 INTERBLOCK SYNC MARK
010110 FORWARD BLOCK MARK 
011010 REVERSE GUARD MARK
001000 LOCK MARK
001000 REVERSE PCC MARK
001000 REVERSE FINAL MARK
001000 REVERSE PRE-FINAL MARK
111000 DATA MARK
 
Hi All;
"" Do you have any good reference to literature on tape read channels? ""
Here is a picture of Schematic for my old MecaAlpha-1 Read and Write section of the Board, it used an LM1458 Op Amp and a 4050 Cmos Ic for this part of the Circuit..
But, I would think You would be better off, using a more modern FPGA type of Chip or like an Arm7, processor that has modern D to A and A to D Sections already in it.. As this schematic is almost as Old school, as what you already have..
If You want more Information on the circuit I have the circuit Description Pages..

001.jpg

THANK YOU Marty
 
Hi All;
Here is a picture of Schematic for my old MecaAlpha-1 Read and Write section of the Board, it used an LM1458 Op Amp and a 4050 Cmos Ic for this part of the Circuit..
Thanks Marrty.

What is a "MecaAlpha-1"? I does look like this device is also just amplifying the read signals. As far as I can see there is no differentiating done when reading data.

But, I would think You would be better off, using a more modern FPGA type of Chip or like an Arm7, processor that has modern D to A and A to D Sections already in it..
Doing DSP operations in the micro controller or host was not what I intended to do when I started this project. But that is also a possible solution to the problem to sample all channels at lets say 1 MHz and then post process it all in the host. It will probably make it possible to create nice algorithms to recover even the most problematic tape.
 
Hi All;
MattisLind, the Meca Alpha-1 is/was an S-100 (Intelligent) Cassette Tape Controller using Phi-Deck Tape Deck's.. The Meca part of it was the Deck and Read/Write Electronics and the Motion Control Electronics.. The Alpha-1 was the Controller Card that plugged into the S-100 Buss.. I have had mine going Partially, but the RF in the air, (Cell Phones and the like) interferes with the Read/Write Signals coming and going from my Tape head, with my Signal levels being so weak..
Therefore I am Quite interested in what You are doing, as a way of Possibly Reading my Old Cassette Tapes and getting the information off of them..
"" As far as I can see there is no differentiating done when reading data. "" Correct, I also would like to pursue Differentiating Ic's for my signal outputs and see if that would Help my reading of the Signal levels any.. Possibly a combination of the two, Op Amp Amplification and Differentiating the signals for the Output from the Board then.. Since I am unable to Read my Tapes, I wouldn't mind changing the Circuits to make the Tapes Readable..
I looked at the Alpha-1 pages and found it did have some sort of Differentiation circuit.. U-47 on the Schematic is (I think) a Mis-print it is a LM1458 instead of a LM1457..

001.jpg 002.jpg

003.jpg 004.jpg

THANK YOU Marty
 
Last edited:
Hi All;
MattisLind, the Meca Alpha-1 is/was an S-100 (Intelligent) Cassette Tape Controller using Phi-Deck Tape Deck's.. The Meca part of it was the Deck and Read/Write Electronics and the Motion Control Electronics.. The Alpha-1 was the Controller Card that plugged into the S-100 Buss.. I have had mine going Partially, but the RF in the air, (Cell Phones and the like) interferes with the Read/Write Signals coming and going from my Tape head, with my Signal levels being so weak..
Therefore I am Quite interested in what You are doing, as a way of Possibly Reading my Old Cassette Tapes and getting the information off of them..

We'll see if I come up with something better. I did some searching on the internet and found this patent filed in 1979 by DEC. They claim it to be much better than all other variants of differentiating and peak detecting circuits. I am going to try it to see if that is any good.

"" As far as I can see there is no differentiating done when reading data. "" Correct, I also would like to pursue Differentiating Ic's for my signal outputs and see if that would Help my reading of the Signal levels any.. Possibly a combination of the two, Op Amp Amplification and Differentiating the signals for the Output from the Board then.. Since I am unable to Read my Tapes, I wouldn't mind changing the Circuits to make the Tapes Readable..
I looked at the Alpha-1 pages and found it did have some sort of Differentiation circuit.. U-47 on the Schematic is (I think) a Mis-print it is a LM1458 instead of a LM1457..
THANK YOU Marty
Now it make sense to me! Indeed the circuit with U47 will do differentiation on the signal.
 
Hi All;
MattisLind, That's funny to me, as it looks exactly like the circuit that Dr. Suding used for His Cassette Read/Write Board, on the Digital Group Computers..

006.jpg 007.jpg

THANK YOU Marty
 
Last edited:
Hi All;
I looked at the Alpha-1 pages and found it did have some sort of Differentiation circuit.. U-47 on the Schematic is (I think) a Mis-print it is a LM1458 instead of a LM1457..

View attachment 22433

THANK YOU Marty

Marty,
To better understand things I am trying to simulate different circuits. I have tested with G888:
0aJSfH1l.png

9rNjlwsl.png


The triangle wave is the approximation of the read signal. The red signal is the output. Not exactly detecting the peaks.

Do you happen to have the values for R35, R38, R39, C22 and C23 for the Meca Alpha interface? It would be very interesting to simulate that circuit as well. What is the intended bit rate of this device?

One general problem is that the noise problem with the more ideal differentiator circuits there is. Therefore I am interested in studying real world examples of working circuits.
 
Hi All;
MattisLind, Wow !!

""Do you happen to have the values for R35, R38, R39, C22 and C23 for the Meca Alpha interface? ""
Yes, C22 is a 470 pf Disk Ceramic, C23 is a .001 mfd Disk Ceramic, R 35 is an 18K 1/4 W, R38 is a 15K 1/4 W and R39 is a 3K 1/4 W..
"" What is the intended bit rate of this device? "" That I could not figure out exactly, But I will give You what it has in the Literature..

010.jpg 005.jpg 006.jpg

Also, Do you have the Values for the OP amp and the Transistor in Your Circuit ??

THANK YOU Marty
 
Last edited:
Back
Top