Page 1 of 1

regulation mod

Posted: 22 Sep 2016, 03:27
by Fonotec
From a .pdf by electronics manufacturer, Analog Devices:

Absolutely not! It is critical to adhere to the absolute maximum ratings in a data sheet to avoid damaging the part permanently. The absolute maximum ratings indicate the limits that a device can tolerate, but not operate at..."

While cloning the ADC-89A8B, by DATEL, I noticed that it uses a comparator, LM710CH, which takes an uneven bi-polar supply. The positive power pin takes +14 V, and the negative power pin takes -7 V. I also noticed that The DATEL 8 bit A/D converter datasheet indicates that its V+ power input should be +15 V (+/- 0.25V). V- is -15 V (+/- 0.25V). Their other converters also only want (uni-, or bi-, polar) 15 V...
DATEL_ADC_PowerReq.jpg (46.17 KiB) Viewed 2239 times
That looks tight, compared to +/- 18 V, which is offered to it by THE LATHE... What's more, the V+ pin of the 8 bit A/D converter "host" is directly connected to the V+ pin of the LM710CH "guest" on the printed circuit board, although that part reportedly wants no more than +14 V on that pin. Clearly, components can tolerate more voltage than the datasheets want us to know... But it seems better not to throw more power at them than is necessary, since their decrepitude is accelerated, otherwise.
LM710H_pinout.jpg (89.82 KiB) Viewed 2239 times
LM710H_AbsMax.jpg (92.25 KiB) Viewed 2239 times
So, the stock LS-76 psu and circuit design is sending the DATEL converters a +/- 18 V supply, which is regulated by two 7818s, in the TO-3 metal can package. Their outputs also go to various other cards in the cage, under the lathe, to power op amps, such as TL081, which have 18 Volts +/- as Absolute Maximum ratings... ...even µA748C, as shown in the datasheet detail, below, (contrary to many µA748 datasheets) may not actually be built quite as impervious to high supplies as the µA748M version... (?)
u748_AbsMax.jpg (155.46 KiB) Viewed 2239 times
...also, this hybrid op amp (HC2000H), below, is powered by the same regulated supplies:
RCA_hybrid_op_amp.jpg (95.59 KiB) Viewed 2103 times
The Depth coil is driven by this hybrid op amp. The power inputs of this part can take bi-polar 37+1/2 VDC for its V+ and V- inputs - a total of 75 V @ 7 Amps.

Since the DATEL converters only want +/- 15V, and the TL081 and µA748 op amps all do well below +/- 18V, I decided to modify the power supply by removing the 7818 regulators and replacing them with 7815s. Updated drawing:
BP_15.jpg (301.78 KiB) Viewed 2239 times


Incidentally, I noticed that the LS-76 manual drawing for the Variable Depth card has an incorrect legend for the power pin of the DATEL Systems, Inc. DAC-98BI module. The legend shown is the symbol for +5V, but the circuit is bringing it in contact with the +"18" V supply. As mentioned, above, it needs to be only + or - 15 V on all of those power pins, however, so, a mod seems in order.
Not+5.jpg (88.98 KiB) Viewed 2239 times


After replacing the two regulators in the psu, I did some tests and found that everything works fine this way, but a recalibration was required to fine tune the fixed pitch, and also the expansions seen under microscope, to agree with the readout on the LPI Display. To confirm that the slightly lower rails were not somehow attenuating the impressing action of the Fixed, or Variable, Depth signals, I also did a test cut of 200 Hz, but with the Left and Right channels 100% out of phase, producing 0 VU signal intensity (ref. +4 dBm @ 1 kHz = 5 cm/sec/channel/diagonal/peak modulation velocity) from the pickup stylus. Indeed, the depth increases by just as much as I told it to, after the regulation mod. The 15 V bi-polar supply is still much bigger than the out of phase automation audio signals going to the hybrid op amp's inputs, so, the attenuated rails are not being exceeded. Therefore, the automation audio would not be using the extra 6 dB of headroom supplied by the original, bi-polar 18 V supply, anyway. Furthermore, the entire Feed and Depth behavior of THE LATHE seems more reliable now. I think it's happy to be less stressed out by the op amp rails... I'll try to keep it this way from now on so that the parts will possibly enjoy a longer life not being under as much pressure.

The Scully family were very smart and made a great lathe, but they were sometimes apparently bored with the mundane concerns of engineering and embraced mirth in their work, as evidenced by a depiction of Kilroy in the drawing for the Display Processor circuit board.

Kilroy.jpg (104.25 KiB) Viewed 2239 times

Re: regulation mod

Posted: 01 Oct 2016, 15:40
by Fonotec
I inquired with DATEL, who are still in business, and they confirmed my concern about bi-polar 18 V power for their converters.

"...the “technical historians” here agree that the power supply specs would be +/- 15V as you indicate. +/- 18V would not have been recommended (or approved had anyone asked). However the op amps inside can generally tolerate higher supply voltages as long as the max between + and – doesn’t exceed 36V. At +/- 18, the unit was running on the ragged edge of the limit..."

I don't know if the person who responded was aware of this, but the converter I asked about has only one op amp - LM710HC - a comparator - and its datasheet specifies there be no more than +14 and - 7 V for its rails. That equals 21 V between + and -...

Re: regulation mod

Posted: 06 Oct 2016, 15:37
by Fonotec
An interesting legend on the PCB of THE LATHE's PSU indicates that at least someone at some time at L. J. Scully felt that the regulated supplies going to the op amps and DATEL converters in the card cage should be bipolar 15, instead of 18, VDC. (The numbers are next to their respective test points.)
Bipolar_15+K.JPG (161 KiB) Viewed 2201 times
The K15 supply is positive and goes to the relays on the Feed Servo board.

The metal heat sink to which the TO-3 package regulators attach, on the back of the PSU, has a +18 and - 18 legend for the two regulators that supply the rails of the op amps and converters.

This heat sink might somehow have been rebadged for 18s from 15s...
Heat_Sink_18s.jpg (270.44 KiB) Viewed 2195 times
(PSU of #656, above)

The font of the number "8." in "18" and "-18," is smaller than that of the other numbers, including "K15" (all the way to the right, if you expand the picture and then drag the frame slider to the right). A change of design occurred, at some stage, after the PCBs, and, possibly, the heat sinks, for the PSUs were manufactured. However, on the PSU of #660, the number, 18, is the same size and font as the other number.

Re: regulation mod

Posted: 03 Apr 2017, 04:19
by Fonotec
I was curious to see how THE LATHE would behave on an actual test cut of music, now that I had recalibrated its feed and expansions after having (apparently, un-)modified THE LATHE's op amp and digital conversion power supplies' regulation (back) to (the original spec (?) of) bi-polar 15 VDC. So, I sneaked down to the studio, this afternoon, and made a test cut of side A of a (ca. 1985) 1/4" 2-trk., dbx Type II NR, 15 ips master tape (titled, Two, as it was the band's second studio recording, following the release of a 7"_33+1/3 RPM EP of the first recording, "Contradiction") of the Cincinnati Punk band, Sluggo, which I had already dubbed, with eq, to a cutting master reel, as in the tradition of RCA's vintage disk mastering. However, I dubbed to a higher fidelity tape format: 1/2", 30 ips, NNR, so, it greatly mitigated the generation loss effects of the dubbing process.

(If I had room in the Neumann SP 272 console frame for four Massivos and four, "8-band" Sontecs, I could have used the original master mix tape for the cutting session and given each track unique parametric eq, while signaling identically, also, in Advance, to THE LATHE, but the acoustics of the cutting room would then be impeded by adding the console furniture needed to hold the necessary, additional, 19" wide, 4RU (high) rack gear in order to make an identical, 8-channel disk mastering signal path. That type of setup entails using identical processors for the cutting amps in an A and B stereo path, and identical processors for the disk computer in the A and B stereo paths. The settings for the first song are made to the A path Program and Advance processors, while the settings for the second song are made to the B path Program and Advance processors. When the first song ends, the console output is cross-faded to the B path. While the second song is being cut, both sets of the A path processors can be adjusted for the logged settings that should be applied to the third song.

One benefit of doing parametric eq in the mastering room down the hall is that its acoustics are superior to that of the cutting room. Part of that is due to the small footprint of the mastering console that only needs to provide one example of each, 19" wide, stereo processor. For complete, 8-channel setups, it seems the best approach would be to use only Danner (or 500-series) cassette format processors, so that the stereo eqs, compressors, limiters, and de-esser, with their three clones, each, don't turn the mastering console into a mixing console. But for aforementioned reasons, it would still be better to proceed with my current, bicameral disk mastering process.)

The MCI JH-110/M is the only advance repro tape machine that provides a full revolution of delay between the arrival of tape at the advance repro head assembly and the program repro head assembly. Therefore, it's the only Advance repro tape deck suitable for use with the LS-76 disk computer, which needs the Advance signals to be 1.8 seconds before modulation at 33+1/3 RPM, and 1.3 seconds before modulation at 45 RPM.

Studer Preview decks usually provide for a 0.6-revolution delay between Preview and Program heads because that's the required delay for the Neumann lathe's disk computer (and also for third-party disk computers that are designed for Neumann lathes).

The LS-76 Feed and Depth automation is based on the last disk computer designed by Capps. It's cognizant of absolute polarity and nestles compatible grooves, sometimes with permissible groove kiss, but consistently avoiding overcuts, when properly calibrated and with appropriate deflections of the automation audio's offset attenuators (if in use). It uses the Left and Right Advance repro signals for computing the Depth automation and uses a copy of the Left Program repro signal, in addition to a copy of the Right Advance repro signal, for computing the Feed automation. (In most cases, Scully says, "Feed" and "Advance," while Neumann says, "Pitch" and "Preview." Scully also uses the term, "Feed nut," where Neumann says, "Half nut.")

For 45 RPM tt speed, which is what the client had requested, at 30 ips tape speed, the distance between the two repro heads, established by the delay path through guide rollers, needs to be 39 inches (for 33+1/3 RPM tt speed at 30 ips tape speed, it would need to be 54 inches of inter-head, tape path distance).

If the 15 ips, 1/4", dbx NR master tape had sounded quite good, I could have used the 1/4" repro head assemblies on the MCI and played the four output channels into two dbx Type II, stereo noise reduction units. The client was not completely happy with the existing sound of the master mixes on tape, however, so I made a master dub on an ATR-102, with all-analog eq (using a Mastering Massivo and a Sontec MES-482-D7 between the ATR capture deck and a Studer A 80 R, which was used for the re-recording session's master tape playback.

Please enjoy this pictorial tale of progress. (...and if you happen to operate an LS-76 with the original disk computer, you can go ahead and perform the modification to the PSU's 18 V regulators, as I did (to bi-polar 15 VDC), since it seems to work fine and ought to provide a margin of safety at least against unnecessary component ageing due to there being (after the regulation mod) less dissipated heat through the two regulators' associated components (especially to protect the DATEL converters, since they're no longer made and are rare in the obso marketplace.)

Note the long and winding road that the 1/2" tape is guided through between the two repro head assemblies. The VU meters show that the Advance heads were beginning to reproduce the stereo recording's intro, while it had not yet reached the Program heads. (There are no Erase, or Record, heads, or circuits, associated with the MCI JH-110/M.)


Advance_Repro_Two.JPG (135.6 KiB) Viewed 1856 times
The test cut, taking place. The base Pitch used was 315 LPI, and the base Width was 2.5 mils. The blank used was a Transco 12" dub. Heat was around 0.5 Amperes. Helium was being streamed through the Ortofon DSS 731 (CD4-capable cutting head), which uses a Transco 329-SH (Ortofon type Micropoint) synthetic ruby.

Cut_Two.JPG (154.89 KiB) Viewed 1856 times
The grooves appear neatly cut, without horns, and with generous land, as the pressing plants tend to like. The three, short-playing Punk songs for each side of Two could easily fit on a 7"_33 RPM record, but a 10"_45 RPM pressing was requested for faster land speed and to be different (and to be more Steam Punk).

Grooves_Two.JPG (209.09 KiB) Viewed 1856 times
Pickup showed good tracking and a decent signal on the VU meters, averaging at around -7 VU, with peaks around -2 VU. I'd probably cut it hotter for the master disk, but, as mentioned, above, today's session was merely a casual, go-, no-go, test cut of just the one side.

The results are in. Regulation mod is a go...