Characteristics of HP-IL Transformers?

[Martin Hepperle]

While I am aware of the PIL-Box (In fact I have one) I am also interested in the original coupling of the HP-IL loop via transformers which seems to put the lowest load on the loop.

I would like to know whether someone has ever characterized the transformers (number of windings, wire diameter, inductance, ...). The documents which I have seen, do not specify any details. Unfortunately I do not have a broken one at had to analyze and I do not want to destroy a working sample.

Modern components in similar applications seem to be Ethernet transformers, but these are rather different and obviously only 1:1, no center taps.

Martin

[cruff]

Have you seen the article in the January 1983 HP Journal? On page 14 they show the transmitter and receiver circuits including the turns ratios for the transformers and in the text discuss the impedance matching with the chips.

[Martin Hepperle]

(08-23-2020 08:37 PM)cruff Wrote:  Have you seen the article in the January 1983 HP Journal? On page 14 they show the transmitter and receiver circuits including the turns ratios for the transformers and in the text discuss the impedance matching with the chips.

Oops, I had forgotten about this article and indeed it contains even the winding ratios and more.
Thank you for this hint! I need to do more reading...

[J-F Garnier]

I don't recall the HP-IL transformer specifications have ever been published or measured.

The turns ratio is known (3:1 for output, about 1:4 for input in two stages) and adapts the +/-1.5V amplitude of the HP-IL signals to the 5V digital levels of the HP-IL chip.

I had the project to experiment with pulse transformers on the PIL-Box, maybe this is what you have in mind too, but didn't do it yet (too many projects...).

J-F

[Hans Brueggemann]

(08-23-2020 07:26 AM)Martin Hepperle Wrote:  I would like to know whether someone has ever characterized the transformers (number of windings, wire diameter, inductance, ...). The documents which I have seen, do not specify any details. Unfortunately I do not have a broken one at had to analyze and I do not want to destroy a working sample.

Modern components in similar applications seem to be Ethernet transformers, but these are rather different and obviously only 1:1, no center taps.

Martin

i could do some inductance measurements on an HPIL transformer.
as for ethernet transformers, have a look at "power over ethernet".

[Martin Hepperle]

(08-26-2020 06:17 AM)Hans Brueggemann Wrote:  i could do some inductance measurements on an HPIL transformer.
as for ethernet transformers, have a look at "power over ethernet".

Hans,

measurements could be useful to support the search for suitable components.

The PoE transformers look promising, but usually seem to be designed for relatively high power loads and higher voltage levels. It is good that they often have center taps resp. multiple coils on a single core. The few I looked at so far (e.g. from Würth Electronics) are relatively large. We don't need power, just voltage ;-)

Martin

[Hans Brueggemann]

yes, only signal transmission, no power needed. bigger cores might facilitate DYI, though. have a look at the inductances.

[Dave Frederickson]

I just noticed that the HP-IL transformer on my 82973A Rev. C is marked:

VALOR PT3493
9100-4226
MEXICO 8832J

I've requested the datasheet and there could be stock here:
https://www.sierraic.com/PT3493

Dave

[Martin Hepperle]

(08-26-2020 06:48 PM)Dave Frederickson Wrote:  I just noticed that the HP-IL transformer on my 82973A Rev. C is marked:

VALOR PT3493
9100-4226
MEXICO 8832J

I've requested the datasheet and there could be stock here:
https://www.sierraic.com/PT3493

Dave

... interesting ... I requested a quote, but In the past most of these "offers" ended up dry in the sand - many of these web sites offer vaporware.
Let's see what happens. At least the datsheet could be interesting.

Hans' measurements seem to be somewhat asymmetric - I had expected the center taped coils to be more symmetric. Or is that a result of measurement + production tolerances?

[Hans Brueggemann]

haha, good catch, martin! they are as symmetric as i can measure them with my RLC meter. note that L = n^2 x AL. hence, a center-tapped winding with pins 1,2,3 and pin 2 being the center tap shows
inductance between 1-2 = L, but
inductance between 1-3 = 4 x L.

btw, according to my scientific findings (i.e., guesstimate the number of turns on the highest inductance...), toroid cores with AL >= 6000 nH will do the trick. will be trying to dig up a useful core later.

[Martin Hepperle]

(08-27-2020 01:54 PM)Martin Hepperle Wrote:  ...
I requested a quote, but In the past most of these "offers" ended up dry in the sand - many of these web sites offer vaporware.
...

They sent me a quote for this "PT3493 Telecom Transformer." The price for one would be 23.17$ (I asked for a lot of 50, 1160$ in total). Shipping withing 3-4 days, so they actually seem to have a stock as listed on their web site. Not really El Cheapo, but also not out of reach.

Did you get the data sheet? Is this really the proper part?

[Martin Hepperle]

(08-27-2020 04:15 PM)Hans Brueggemann Wrote:  haha, good catch, martin! they are as symmetric as i can measure them with my RLC meter. note that L = n^2 x AL. hence, a center-tapped winding with pins 1,2,3 and pin 2 being the center tap shows
inductance between 1-2 = L, but
inductance between 1-3 = 4 x L.

btw, according to my scientific findings (i.e., guesstimate the number of turns on the highest inductance...), toroid cores with AL >= 6000 nH will do the trick. will be trying to dig up a useful core later.

Hans,

thank you for the explanation - I have to confess that I am an engineer, but not in the field of electronics.
Especially the world of analog electronics is difficult to understand and probably full of magic :-)

Martin

[Hans Brueggemann]

tbh, to me as an enthusiastic armchair avionist,
airfoils are encompassed by the very same aura ;o)

[Hans Brueggemann]

seems that my MK.1 eyeball needs some serious re-adjustment, as the best toroid core i could find wrt AL-factor is a TDK B64290P0037X038 at just AL=2530 nH, material grade T38 (there is another one with AL = 3600 nH, but basically unobtainable in single-digit figures).
the core can be ordered at digikey for a whopping 0.43€/ea.. three cores are required to make an HPIL transformer assembly.

for the given inductances from my previous post and the turns ratios laid out in the HP journal january 1983, p. 14, the number of turns would be as follows, where the first (pin-)number of the winding indicates the start of the winding, indicated by a small dot in the diagram:

core 1
winding 02-01: 48
winding 16-15: 12
winding 15-14: 12

core 2
winding 05-04: 103
winding 06-04: 103
winding 13-12: 24

core 3
winding 08-07: 99
winding 10-09: 33

for the windings, enamelled copper wire of 0.11 mm - 0.20 mm dia. (AWG38 - AWG32) should fit. AWG32 is already a tad on the "dense" side @ about 0.76 packing factor, which means that windings 05 and 06 should be done with extra care.

[Martin Hepperle]

@Hans:
That sounds very promising. Do you think the cores must be of the toroid shape? Manually threading 100-200 turns through them does not sound like a fun evening... except there is some magic involved.
Could they not be replaced by a straight or a split core type which allows for easier winding?

@Dave:
Did your receive the datasheet? Would be interesting for comparison.

[Hans Brueggemann]

the TDK B64290P0037X038 basically allows you to have the complete HPIL transformer occupy the same "real estate" as the original one. i'm thinking of something like a milled DIL-16 IC socket with the three cores glued on top of it.
however, if your pcb space permits, you can of course go with larger toroids, such as TDK B64290L0062X046, for example. on top of being easier to handle, it has a much higher AL-factor of 6400 nH, with allows you to reduce all winding numbers by a factor of approx. 2.53 . for example, winding 5-4 would be reduced to 41 turns.

the other option that you probably had in mind is using a split-core transformer assembly, which in fact is very easy to wind. but such a contraption can become "pretty large" and is much more expensive. the reason is that you need a "set" of two core halves with high-precision planed faces (we don't need an air gap), a bobbin, and a set of clips to hold everything together. also, we need 3 sets in total to form an HPIL transformer. have a look at this core-set:
RM4 low profile gap-less
the correct ordering numbers for clips and bobbin can be found in that datasheet.
... but then again, we are back at base one ;o)

[Martin Hepperle]

I fear that a commercial quote for winding toroid cores would require an order of 100'000 and up, just for considering to set up a production run. Manual winding looks at least like a "very interesting" job.

For manual winding the split cores look much more attractive to me.

If I read the accessory list correctly, the simple split core variant with the required parts (bobbin, covers, clamp, no adjustment screw) would cost maybe 3€ for one, respectively around 10€ per transformer with 3 of them. This would be acceptable for me.

Also, for me, size would not be so important, if the end result still fits into something like a PIL box. I would not want to recreate a HP-41/71/75 internal HP-IL module, but would like to have a interface building block (hard- and firmware) for microprocessors similar to what the original combination with the HP-IL chip offered.
Similar to the PIL-box but with HP-IL frame de/encoding in the same hard/firmware so that no extra laptop or Arduino etc. is needed. Maybe with an ESP32 or similar device.

If such an interface is desired for something small, e.g. as a serial to HP-IL translator for a Swiss Micros calculator, it would have to go into an external box.

[Dave Frederickson]

(09-02-2020 01:03 PM)Martin Hepperle Wrote:  @Dave:
Did your receive the datasheet? Would be interesting for comparison.

Alas, no.

From what I can find, Valor was acquired by Pulse Engineering in 1998. Pulse Engineering's world headquarters was in my zip code. The signage along the road disappeared 5-10 years ago, but I think there's still a sales office nearby. Want me to go knock on their door?

[Martin Hepperle]

(09-03-2020 02:09 PM)Dave Frederickson Wrote:  ...
From what I can find, Valor was acquired by Pulse Engineering in 1998. Pulse Engineering's world headquarters was in my zip code. The signage along the road disappeared 5-10 years ago, but I think there's still a sales office nearby. Want me to go knock on their door?

Probably not worth the hassle - if they changed ownership, they probably shed their old identity and knowledge w.r.t. this old stuff.

I am currently looking into datasheets for ISDN telecom transformers - there seem to be many with multiple center tap coils.

[J-F Garnier]

(09-03-2020 11:14 AM)Martin Hepperle Wrote:  I would not want to recreate a HP-41/71/75 internal HP-IL module, but would like to have a interface building block (hard- and firmware) for microprocessors similar to what the original combination with the HP-IL chip offered.
Similar to the PIL-box but with HP-IL frame de/encoding in the same hard/firmware so that no extra laptop or Arduino etc. is needed. Maybe with an ESP32 or similar device.

The input path uses two transformers in cascade, this was designed in this way for the ESD protection.
If you plan to use pulse transformers with something else than the HP 1LB3 (and later versions), you can use only a single transformer for each input/output path. I would not recommend to change the design for the 1LB3 because the chip ESD protection is special and may depend on the original arrangement (although early HP products/prototypes used discrete pulse transformers, see here for instance).

With the two transformers in cascade (input path), the effective inductance from the HP-IL side is reduced by a factor of ~2 and is about the same as the inductance (also from HP-IL side) of the output path (~2.5mH). So you can take the output transformer characteristics as a model.

J-F