• Please review our updated Terms and Rules here

D662/D664 Diode Replacements?

vrs42

vrs42

Veteran Member
Joined
Nov 24, 2009
Messages
1,003
Location
Beaverton, Oregon
It has happened to me a couple of times now. I go to Mouser, and I spend an afternoon staring at data sheets for switching diodes. I'm looking for diodes with low and high forward voltage drops, to replace standard bits of the logic in Bxxx, Rxxx, and Sxxx logic gates.

There are almost always two types used:
D664 is used to AND inputs without shorting them, and ideally would drop as little voltage as possible. (If they'd been readily available, they'd probably use a Schottky diode.)

D662 is used to form an inexpensive voltage regulator. Four such diodes are used in series, and their cumulative drop is expected to be 3V. The other 12V (out of 15) is dropped across a 1500 ohm resistor, which also limits the current to about 8 mA. For this to work, the forward drop needs to be about 0.75V per diode at that current. The diodes will quickly shunt less current if the load shifts.

Anyway, my diodes arrived today, and I built a regulator as described above and...the voltage drop is 2.88V regardless of which of my carefully selected diodes is used. Admittedly, an error margin of 4% or so isn't terrible, but it isn't great either, and I was also hoping for a lower value on my D664 replacements. (The diodes I ordered this time are 1N914 and 1N916 with specific revision numbers mentioned on the data sheets, but seem to behave identically regardless.)

Has anyone had any luck or care to chime in on what best to replace these diodes with? Ultimately I'm trying to fabricate some missing modules for my Logic Lab. The issue would also be relevant for straight-8 or 8/S repair, as well as various peripheral gear from the Negibus era.

Thanks,

Vince
 
thunter0512 said:
Use one 3V Zener instead of 4 D662?
Click to expand...
Well, that wouldn't be following the original designs, which was one of my goals. I don't want to redesign the Logic Lab (or other vintage gear); just replace the missing and defective bits.

To be functional, I believe a zener based regulator would need the output filtered, as the diode itself would not be fast enought to respond to logic-speed load transients. Then I'm effectively replacing the diodes with relatively expensive caps (and a more expensive diode).

Although, for another project (involving logic level conversion), I'd love to see 3V zeners that could run at switching diode speeds.
 
You could add around 0.2 to 0.3V with an additional Schottky diode in series with the 1N914's to get it closer to 3V, try a BAT 46. Might also suit the other application.

www.vishay.com/docs/85662/bat46.pdf

If you needed a lower capacitance faster Schottky signal diode (though these have more fwd drop) for the AND application (though diodes make good OR gates), use a 1N5711, rather than the BAT46.

You can adjust the diode current (within reason) to get the fwd drop closer to your target.

A few series diodes do not make for a wonderful voltage reference because of the negative temperature coefficient, unless perhaps they are used to cancel a positive tempco in the circuit elsewhere.
 
Last edited:
Hugo Holden said:
You could add around 0.2 to 0.3V with an additional Schottky diode in series with the 1N914's to get it closer to 3V, try a BAT 46. Might also suit the other application.

www.vishay.com/docs/85662/bat46.pdf

If you needed a lower capacitance faster Schottky signal diode (though these have more fwd drop) for the AND application (though diodes make good OR gates), use a 1N5711, rather than the BAT46.

You can adjust the diode current (within reason) to get the fwd drop closer to your target.

A few series diodes do not make for a wonderful voltage reference because of the negative temperature coefficient, unless perhaps they are used to cancel a positive tempco in the circuit elsewhere.
Click to expand...
I like that idea. Though I'm not sure how authentic to the original design, I have to admit I don't fully grasp what the D664 diode was or what the modern equivalent is. Perhaps it is simplest to think of it as an approximation of a Schottky diode.

A clarification of the actual use of the "regulator", is that is it used primarily to clamp inputs or outputs that might otherwise exceed the logic "low" value of -3V. The series diodes are connected to -15V via the 1500 ohm resistor. Individual logic levels that need to be clamped do get their own D664, attached to the -3V clamping rail provided by the four D662. This, I think closely imitates what you are proposing. (There's no concern about voltages above 0V: that would be an unrelated clamping circuit.)

The design allows DEC to control the speed, etc. by manipulating drive current levels, without worrying that the output voltages will swing outside the 0V to -3V range. Any extra current flows will be routed through the diodes.

I'm often confused by the terms OR and AND when they are used here. As in positive logic, the ground level 0V is the one things are pulled to to merge them. The "non-asserted" level is then the -3V. I guess calling it OR or AND is meant to signify the designer's intent. These are general purpose logic modules, so calling them NAND or NOR will depend on polarity of the inputs in the application.
 
I've been using 1N4153 for D664. 1N3604 was listed as cross for D664 and 1N4153 is cross for 1N3604. If you wish me to measure some characteristic on those I can. The only time I've had to replace D662 is a TU55 that had a large number of dead components so likely power hooked up wrong before I got it. I had 1N645 diodes around which is listed as the cross. A couple of the flip chip black modules were bad. I tested one board which I replaced module with 4 1N645 diodes and it measures almost 3V.
 
djg said:
I've been using 1N4153 for D664.
Click to expand...
I was able to find a datasheet that quotes the forward voltage for 1N4153 at 1 mA (typical logic input current) as 0.59V (0.62 for 2 mA). That same sheet quotes 1N914B and 1N916B as 0.72V and 0.73Vat 5 mA:
Unfortunately, the matrix is quite sparse for most of the diodes.

The other problem is that these old data sheets don't seem to be a good match to the new parts I get these days from suppliers. (Nor, for that matter, do the parts seem to match the suppliers current datasheets, at least for forward voltage drop.)

Good to see the 1N645 voltage come out about right. Might want to check it's reaction time, though, since it seems to be a 20 pF thing.
 
Have this document from ages ago ...
Code: 
    DEC                EIE
D-003, D003            1N994
D-662, D662            1N645
D-664, D664            1N3606
D-668, D668            Two 1N3606 in series
D-670, D670            1N3653
D-671, D671            1N3653
D-672, D672            1N3653
DEC 999, DEC999        MM999
DEC 1008, DEC1008      MM1008
DEC 2219, DEC2219      2N2219
2N2904                 2N2118A
DEC 2904, DEC2904      2N1132
DEC 3009, DEC3009      2N3009
DEC 3009A, DEC3009A    2N3009
DEC 3009B, DEC3009B    2N3009
DEC 3500, DEC3500      2N3500
DEC 3568, DEC3568      2N3568
DEC 3634, DEC3634      2N3634
DEC 3638, DEC3638      2N3638
DEC 3639, DEC3639      2N3639
DEC 3639B, DEC3639B    2N3639
DEC 3639-2, DEC3639-2  2N3639-2
DEC 3715, DEC3715      2N3715
DEC 3790, DEC3790      2N3790
DEC 4258, DEC4258      2N4258
DEC 6531, DEC6531      MPS6531
DEC 6534, DEC6534      MPS6534
SDA-6, SDA6            2N2060
1N429                  1N429 6.2V 5%
1N449                  1N449 6.2V 5%
1N762                  1N762 5.5V 250 ohms 5%
1/4M6.8AZ5             1N4099
1/4M2.4VAZ5            1N4370
MR2064                 1N4001
MR2066                 1N4003
that gives equivalents for the D662 and D664.
Sometimes DEC would use the standard parts but with an additional screen for specific characteristics.

These guys https://www.littlediode.com/components/home.php have both in stock for GBP 3.99 (US$5.07) ***EACH***. Yikes.

Better price here for the D662 1N645 at Jameco:
https://www.jameco.com/z/1N645-Majo...Diode-1N645-225-Volt-400mA-DO-35_2303255.html
at US$0.79 each.

Or here for the D662 1N645-1 at Digikey:
for US$1.96 each for MicroChip as the vendor, it is a current production part.
 
Last edited:
You really should be cautious here. A lot of parts have infiltrated suppliers that are not genuine, and some suppliers are not stocking suppliers and they acquire the part or label an alternative.

If I were searching for genuine 1N645 diodes, I would buy them from Surplus Sales Nebraska, they even have the mil-spec version, and they are all original NOS parts. and for less than 50c
 
Hugo Holden said:
You really should be cautious here. A lot of parts have infiltrated suppliers that are not genuine, and some suppliers are not stocking suppliers and they acquire the part or label an alternative.

If I were searching for genuine 1N645 diodes, I would buy them from Surplus Sales Nebraska, they even have the mil-spec version, and they are all original NOS parts. and for less than 50c
Click to expand...

The 1N645-1 at DigiKey is a current production part from MicroChip (MicroSemi subdivision), so its lineage is not an issue.

Or 1N645 from NTE at DigiKey: https://www.digikey.com/en/products/detail/nte-electronics,-inc/1N645/11644785 for US$2.22 each.

The 1N3606 however appears to be obsolete/no longer produced.
 
You must log in or register to reply here.
Back
Top