Probably the most sophisticated CP/M-80 machine ever made

[precaud]

The boot screen looks innocent enough, though that "g" in the BIOS version has special meaning...

[precaud]

Front and rear views suggest this isn't a typical Z80 machine.

[precaud]

And though you can run the generic CP/M software of the day on it, it was really built to do things like this.
1st pic is anechoic magnitude and phase vs frequency of a tweeter on a test baffle.
2nd pic is an 3D "waterfall" display (energy vs time vs frequency) of a speaker system.

[RoGeorge]

I'm intrigued by the Charge/Battery indicators, thought I see a power cord in the pics.

Is that portable, can it run on batteries only? 

[precaud]

So what is it, Johnnie?

It's a Tecron TEF-10 TDS Analyzer. A purpose-built Z-80 CP/M-based computer for doing acoustical measurements using Richard Heyser's Time Delay Spectrometry techniques. Inside are three Z80's, an AMD 9511 FPU, the usual 64kB of ram, sine and cosine oscillators, balanced mic and line-level preamps, etc. The "g" in the Bios string signifies support for the GPIB interface.

I bought this new in 1983. It has been in storage since 1989, when I upgraded to it's big brother TEF12+.

I set it up yesterday and it booted up. It ran all the software (from 38-yo 5.25" floppies) just fine. Time in a bottle.

I'm thinking of setting up a simple rig to measure headphones. My AKG C451E with CK2 capsule from the same period also appears to be just fine.

Fun stuff.

[precaud]

Yes, it can be powered from a 12V car battery. It has an internal NiCad pack that powers it for 10-12 minutes in the event of a power outage, giving you time to save your data and shut down gracefully.

[RoGeorge]

Wow, impressive!  Now I notice, indeed, that instead of ON/OFF the switch on the back reads DC/AC.

[precaud]

Yes, it's an impressive machine, the brain child of Crown's Gerald Stanley and his crew. Software was written in Kaypro's S-BASIC and assembler. Keith Jebelian did most of the OS and IO hardware code. Even with GPIB support added, it still had an impressive 57kB TPA!

[precaud]

The only thing that gives me the willies about using it now are the 720kB 5.25" floppies. I wish there was as SSD or flash replacement for 'em.

[Gyro]

The drives themselves are pretty bulletproof, or at least serviceable/repairable (much more so than 3 1/2"). You just need to stock up on media and make lots of copies!

[BrokenYugo]

The only thing that gives me the willies about using it now are the 720kB 5.25" floppies. I wish there was as SSD or flash replacement for 'em.

I thought that was an available product, "floppy emulator", pulls images off a USB drive and acts like a floppy drive on the other end.

[precaud]

You just need to stock up on media and make lots of copies!

Yeah, the media's the problem.

I thought that was an available product, "floppy emulator", pulls images off a USB drive and acts like a floppy drive on the other end.

The only ones I've seen are for 3.5", 1.44M PC format.

[gf]

Regarding emulator, the important question is, which hardware interface to the floppy drive is supported by the computer? I guess it is not yet ATA?

[precaud]

Definitely pre-ATA... I think most CP/M machines used a Shugart-standard interface, no?

[RoGeorge]

stock up on media and make lots of copies!

That's generally a good advice, yet I have 5.25'' floppy disks (same as old as from 1985) and still working OK (only rarely used).

For a backup, I've seen many projects for floppy disk unit (FDU) emulators based either on HDD, or on an SD card + microcontroller acting like a FDU.  The FDU interface stayed the same, so a whole unit can be replaced now with an emulator.

For a backup, either find an old PC motherboard and install an old Linux that can still read 720kB floppy disks, connect the CP/M unit to the IBM/PC and make a dd backup image of the entire disk(s) with the help of the PC.

If you don't find any, there was a tool named Kermit, for CP/M, tool that was able to exchange files over the RS232 port, or you can try to write one if the CP/M disks have some OS tools for a programming language.

Another option would be to search for existing backups on the Internet, maybe somebody else already copied the disks and uploaded them somewhere.  Last time I looked there were many places preserving CP/M software.

With the current PC speeds it would even be possible to record the row pulses at the output of the floppy head, the RDATA line. 

Also, there might be EPROM chips (UV erasable) that would worth a backup, too, but only if they are sitting on a socket.

[SiliconWizard]

Definitely pre-ATA... I think most CP/M machines used a Shugart-standard interface, no?

Yes.

[precaud]

... I have 5.25'' floppy disks (some as old as from 1985) and still working OK (only rarely used).

Same here. It is quite possible I am 'misoverestimating' (thank you GWB for bringing us that elegant piece of Amerenglish ) the potential unreliability of the floppy system. It never gave me problems.

For a backup, I've seen many projects for floppy disk unit (FDU) emulators...

For a backup, either find an old PC motherboard...

If you don't find any, there was a tool named Kermit, for CP/M...

Another option would be to search for existing backups on the Internet...

Thanks, but none of those things are necessary, I still have the master discs, and I can get files on or off the computer with ease by several means. It came with a utility that turns drive B into an MSDOS drive for read, write, even format.

Perhaps the best thing to do for now is to see how many good 2S2D disks I have, make working backups, and just use the thing. It's not like I'll be making hundreds or thousands of measurements. Dozens, maybe. I have several boxes of NOS 5.25" HD floppies but have already discovered they don't format well in DD drives. Something about write current being too low.

[RoGeorge]

1st pic is anechoic magnitude and phase vs frequency of a tweeter on a test baffle.
2nd pic is an 3D "waterfall" display (energy vs time vs frequency) of a speaker system.

No idea what methods are used to measure those in the audio industry, so the ways I'll try to measure such params would be one of these:
- either slowly sweep the frequency with a sinusoidal signal while continually measuring the amplitude and phase relative to the source signal (so analog phase and amplitude measurements logged while sweeping the frequency)
- or send a single square pulse, digitize the response waveform with an ADC, then turn the recorded time-domain response into a frequency-domain response with the help of Fourier

For either method, I think a single output signal should be enough, but I've seen on the back panel both sin and cos outputs, so I wonder why does the instrument needs both.

Do you have any hints how the instrument measures, please, or why does it has quadrature outputs? 

[PKTKS]

The only thing that gives me the willies about using it now are the 720kB 5.25" floppies. I wish there was as SSD or flash replacement for 'em.

It is possible...  I have seen some folks adapting an interface for SD-CARDS
using a custom based I/O attachment fixture... even possible it they (already have)
are also capable to offer S-100 (Altair 8800)....

Although I have seen this for sale  for PCs the Z80 I/O is very close to 8085/8080 if not identical in this particular case..  someone with will and time should be able to interface a Z80...

Those graphics written in Assembly are a work of art..

Paul

[precaud]

No idea what methods are used to measure those in the audio industry, so the ways I'll try to measure such params would be one of these:
- either slowly sweep the frequency with a sinusoidal signal while continually measuring the amplitude and phase relative to the source signal (so analog phase and amplitude measurements logged while sweeping the frequency)
- or send a single square pulse, digitize the response waveform with an ADC, then turn the recorded time-domain response into a frequency-domain response with the help of Fourier

For either method, I think a single output signal should be enough, but I've seen on the back panel both sin and cos outputs, so I wonder why does the instrument needs both.

Do you have any hints how the instrument measures, please, or why does it has quadrature outputs?

Personally, for speakers and acoustics, I never had a need for the quadrature output. My guess is it was provided for completeness. The digital version of the signal is used for computation internally.

In simplest terms, TDS is a swept sinewave with a phaselocked swept bandpass receiver with programmable bandwidth and delay. Removing the airpath delay allows it to measure absolute phase. And at a given sweep rate, varying the bandwidth of the receiver IF controls the "time window" of the measurements; narrower makes then anechoic, wider to include room reflections. With a directional microphone, you can isolate and measure the spectra of individual reflections to the complete exclusion of the direct signal. Or absorption coefficients of materials placed over the reflected surfaces. All this with the the huge SNR advantage that swept measurements give over FFT-based impulse or chirp techniques. And there's a whole time domain aspect to it as well; an FFT of the IF signal yields the energy density vs time.

The best thing to do is to read up on the basics of TDS. Google "Time Delay Spectrometry" and you'll find plenty of material. Heyser's AES papers are highly recommended.

[precaud]

It is possible...  I have seen some folks adapting an interface for SD-CARDS
using a custom based I/O attachment fixture... even possible it they (already have)
are also capable to offer S-100 (Altair 8800)....

Although I have seen this for sale  for PCs the Z80 I/O is very close to 8085/8080 if not identical in this particular case..  someone with will and time should be able to interface a Z80...

Will and time, that is the key...

Those graphics written in Assembly are a work of art..

Totally agree!

[PKTKS]

It is possible...  I have seen some folks adapting an interface for SD-CARDS
using a custom based I/O attachment fixture... even possible it they (already have)
are also capable to offer S-100 (Altair 8800)....

Although I have seen this for sale  for PCs the Z80 I/O is very close to 8085/8080 if not identical in this particular case..  someone with will and time should be able to interface a Z80...

Will and time, that is the key...

But even putting the fixture in place...

The CP/M kernel was never meant to use any other DEVICE than those at default installation...

Finding a suitable format  on an alien DEVICE like an SD-CARD will certainly be a problem...

Paul

[precaud]

That is my sense of it as well. A BIOS rewrite would be required.

[jmelson]

The only thing that gives me the willies about using it now are the 720kB 5.25" floppies. I wish there was as SSD or flash replacement for 'em.

I have a Pick and Place machine that runs Windows 95, and used a 3.25" floppy for data exchange.  I got a device mostly sold for use with music synthesizers that makes a USB memory stick (thumb drive) look like a bunch of 1.44 MB floppies.  I'm not sure that would work for your 720KB disks, but maybe somebody makes one of these that will do 720K.  What I have is the Gotek-SFR1M44-U100.

Jon

[precaud]

I have a Pick and Place machine that runs Windows 95, and used a 3.25" floppy for data exchange.  I got a device mostly sold for use with music synthesizers that makes a USB memory stick (thumb drive) look like a bunch of 1.44 MB floppies.  I'm not sure that would work for your 720KB disks, but maybe somebody makes one of these that will do 720K.  What I have is the Gotek-SFR1M44-U100.

If the interface was the same, that could be an option. This https://www.vcfed.org/forum/forum/genres/other/65362-identifying-shugart-interface-5-25-floppy-drives?p=794197#post794197 discussion on vcfed.org suggests it's the same at the drive end but not at the controller end. They're different connectors at the drive end, elso, though I seem to recall seeing adapters. It's been a long time since I last futzed with this stuff.

[bw2341]

I haven't used floppies in a while, but I do follow a few retrocomputing YouTubers for entertainment.

One thing to watch out for is mould on the disk surface. It can cause the magnetic coating to flake off and contaminate the heads and other disks. If you see a white deposit that doesn't wipe off on the disk, you need to discard it.

The 720KB floppies are probably a rare 80 track format (called quad density?) using a media similar to 360KB DSDD floppies. The HD 1.2MB floppies are also 80 track but use 15 sectors instead of 9. They use a cobalt based formulation with a higher coercivity which is probably why they don't work in your drive.

Here's a huge webpage with more info:
https://www.retrotechnology.com/herbs_stuff/drive.html

The electrical interface of floppy drives are almost always the Shugart pinout so an emulator should work. Gotek seem to be an inexpensive design that has been repeatedly cloned by Chinese manufacturers. I also found a open project from a French developer with an official manufacturer and seller in Poland. They seem to have a large list of known working equipment including test instruments.

http://hxc2001.free.fr/floppy_drive_emulator/

[precaud]

Very interesting links, bw2341, thanks. I'll look into them.

Yes, these are "quad density" drives, but use double-density floppies. I remember that being a point of confusion (and uncertainty) back in the day. Mould is rarely an issue here in the desert southwest.

I found an almost-full box of brand new 2S2D floppies in a filing cabinet last night. This should be plenty for my needs.

[precaud]

Some random thoughts, and items of historical interest.

I was showing this to my son last night. He is a professional programmer with focus on UI, and was 3 when I bought the TEF10. Watching this work in a 64KB memory space elicited blank stares; "I don’t know of a single device I own that could run on even 100 times that much memory."

It's amazing what can be done in a 64k memory space if you set your mind(s) to it.

In use, there is a lot of disk activity, swapping program modules in and out. Really slows things down. This thing would benefit hugely from an SSD.

Some historical context: When the TEF10 was released, the IBM PC was already out and taking the world by storm. As you can imagine, Tecron took a lot of flack for coming out with a Z80 CP/M-based system, especially considering the price point. So why did they?
: You can't keep, or control, time accurately with an 8086/8088, because of the prefetch buffer (called pipelining back then); instructions don't always execute in exactly the same number of clock cycles. But you can with the Z80.
: Also, as I understand it, development of the system started in 1979, well before the PC was a "thing".

Not many TEF-10's still exist, because Tecron made a concerted effort to remove them from the market. They didn't want to (or couldn't guarantee being able to) service them. They did this by offering a generous tradein allowance to owners to upgrade to a TEF12, and later, TEF20. (I believe availability of the 720k floppy drives played a major part in this; they switched to the 1.2MB drives in the TEF12).

A note on serial numbers. The serial number on the back of the unit (22288) is not exclusive to this product. The only estimate I've heard is, they made around 175 TEF-10's. That number seems low to me. The only way to know, and to know where a particular unit falls in the sequence of release, is to look at the last digits of the CP/M license number, which were issued in order and written on the software distribution disks that came with the unit. This unit is #00007, one of the "pre-production" units offered.

[bw2341]

Via Adrian's Digital Basement, I found a collection of Shugart service manuals. I think the format of your drives would correspond to the SA465.

http://www.bitsavers.org/pdf/shugart/SA4xx/

As far as I can tell, the pinouts of 5.25 and 3.5 inch floppy drives are the same. The connector changed from PCB edge card to dual row 0.1 inch pin headers. If you want to connect a modern emulator to your floppy controller, you can even snap an IDC dual row connector directly onto your drive cable.

There are also non-permanent edge card to dual row adapters available. I found this from searching 3.5 to 5.25 floppy cable adapter:

https://www.ebay.com/itm/124055375980

Here's a PCB design to make one yourself:

https://www.pcbway.com/project/shareproject/5_25__Floppy_edge_connector_to_34_pin_3_5__floppy_drive_adapter.html

As you can see, the pins are connected straight through.

Even USB floppy drives use a Shugart interface internally. Here's Limor "Ladyada" Fried hacking a USB floppy drive via the Shugart interface and a link to a person who figured out the pinouts on the 1mm pitch flex cable connector.


https://jestineyong.com/examining-and-testing-a-small-universal-fdc-to-usb-adapter/

So for your test equipment, the first step would be figuring out if your drives have a standard Shugart interface and connector. If so, you can get an adapter to connect an emulator. The Gotek seems to have limited manufacturer support, but retrocomputing enthusiasts have hacked it with alternate firmware to support different formats. The HxC seems to have a discussion forum, so more direct support might be possible.

[precaud]

Dear bw2341,
You're making it very difficult to take the lazy route here    Thanks for your good sleuthing   
I hauled out the schematics (yes, I got them to send me a full set of schematics back then...). It uses a WD1797, which was a fairly common floppy disk controller. The drives are not specified, but they are Mitsubishi. I tried removing them but no luck. The bottom mounting screws are inaccessible without a significant disassembly of half of the unit. Perhaps a downside of having a pre-production unit ... sigh      I'm not up for doing that right now.

[RoGeorge]

Not sure why dismantling the 5.25 FDUs, do you plan to replace them with 3.5?

In my experience the 3.5 were less reliable, and the DD (720kB) 3.5'' were rather rare even back then, when compared to the 1.44MB ones.

The FDU pinout for 3, 5 or even the 8'' was the same IIRC, only the connectors were different.  The few signals needed, Step Dir and Side Select for heads, Read, Write, Idx, Motor on, etc, are well known and documented, there is no need to reverse engineer that or the pinout.

[precaud]

No, not "dismantling the drives"; dismantling the internals so the drives can be removed, to see the model number, to clean the heads, replace them, whatever.

[m k]

The interface is actually pretty simple and the drive is merely a spinner.

There are two possibilities, a track write or a sector write, read is trivial.
Today's MCUs have no problems timing those, no matter what.
Only possible problem is unknown sector interleaving but it can be analyzed when formatting happens.

Since a floppy is so small a thumb drive can hold a library of them.
Surely there is also an emulator with a display and some buttons.

[Deni]

In use, there is a lot of disk activity, swapping program modules in and out. Really slows things down. This thing would benefit hugely from an SSD.

These USB floppy emulators won't be any faster, since they emulate physical drive mechanics in order to fool floppy disk controller...

[precaud]

These USB floppy emulators won't be any faster, since they emulate physical drive mechanics in order to fool floppy disk controller...

Well if that's the case, I won't bother with them now. Thanks for the headsup.