Hi All
After quite a few years in planning and one year in the making, a modern replacement for DEC's H7864-A and H7864-B power supplies for BA23 enclosures is now available.
A replacement for DEC p/n 30-21749-01 (Astec AA12131), it is fondly named H7864-C.
The power supply is based around the following main components:
- A metal shell, 1:1 copy of the original in shape and size. Even the color is right.
- Two TDK/Lambda LS200-L power supplies, these are 200W power supplies - allowing for 5V @ 40A and 12V @16.7A total loads
- A custom PCB, generating the POK and DCOK signals and the fan supply voltage
- All the various connectors and looms needed to make it all work and attach directly to any existing BA23 without modification
The custom PCB has several assembly options:
- Fully digital - a Microchip PIC microcontroller generates the POK and DCOK signals and a fake 50Hz LTC signal
- Fully analog - 50/60Hz line-connected circuitry samples the AC supply and generates the LTC signal, and an analog delay circuit provides the POK and DCOK signals using a 74AC14
- Mixed digital + analog - the PIC microcontroller generates all the signals while following the line 50/60Hz in order to generate the correct timing POK and DCOK signals upon a power failure
All assembly versions support:
- Jumper-selectable fan speed (10V or 12V)
- Jumper-selectable LTC line enable/disable and pull-up option - in case anyone needs it
- On-board LEDs for 12V, 5V, POK and DCOK state indication
The only difference from the original is the remote on/off mechanism. The original would have a switch connected to the remote on/off input for enable/disable functionality. The H7864-C needs a positive voltage to (remotely) disable the supply.
Otherwise the H7864-C exceeds its predecessors in total load capability, and in expected lifetime - much better than a used 30-year-old unit anyway.
Tested and working now in a couple of BA23's for several weeks. No issues at all - as expected.
I would still need someone with a PDP11 verify that the LTC signal performs as expected. But that would mean someone with a PDP11 first needs to buy one.
Finally, some observations I can provide, for those of you who may wish to undertake a similar project for a DEC power supply:
1. It is a lot more work than you think
2. Prepare for some serious and tedious re[verse]-[re]engineering work
3. It is more expensive than you think
By the time you have got yourself a set-up, all the required parts, metalwork, connectors, CAD software, prototype PCBs, crimping tools, soldering equipment, pins and what have you, you would have spent a lot of money.
4. It may well end up being less than perfect. One issue I have is the fan connectors - the old DEC fan connector seems to have its two female sockets closer to its surface than the presumably identical connector you can get today. It means the new connector barely holds to the fan connector.
Some gentle filing is necessary, to shave off enough PVC (or whatever) to make the new connector fit nicely on a fan and hold its position.
I will include several photos of a finished unit, inside and out, in the next post. I will also post the PCB schematics and some oscilloscope snapshots showing tests of the PIC microcontrollers' DCOK and POK response to a power failure (in the mixed analog-digital option). I did these tests in order to verify that the timings are within the required 4ms limit. If you do not run a mission-critical application, you don't really need this option.
Now about purchasing one - yes I can build one for you. Please contact me directly. Currently all existing PCBs are built for digital option - as I think no one really needs anything else.
The H7864-C is a fully functional, commercial-grade product but it is intended for hobbyist use. It will probably pass UL and FCC certification without a problem, but this is something I will leave for when an order for 100 units comes along. Which will never happen.
-Alon.
After quite a few years in planning and one year in the making, a modern replacement for DEC's H7864-A and H7864-B power supplies for BA23 enclosures is now available.
A replacement for DEC p/n 30-21749-01 (Astec AA12131), it is fondly named H7864-C.
The power supply is based around the following main components:
- A metal shell, 1:1 copy of the original in shape and size. Even the color is right.
- Two TDK/Lambda LS200-L power supplies, these are 200W power supplies - allowing for 5V @ 40A and 12V @16.7A total loads
- A custom PCB, generating the POK and DCOK signals and the fan supply voltage
- All the various connectors and looms needed to make it all work and attach directly to any existing BA23 without modification
The custom PCB has several assembly options:
- Fully digital - a Microchip PIC microcontroller generates the POK and DCOK signals and a fake 50Hz LTC signal
- Fully analog - 50/60Hz line-connected circuitry samples the AC supply and generates the LTC signal, and an analog delay circuit provides the POK and DCOK signals using a 74AC14
- Mixed digital + analog - the PIC microcontroller generates all the signals while following the line 50/60Hz in order to generate the correct timing POK and DCOK signals upon a power failure
All assembly versions support:
- Jumper-selectable fan speed (10V or 12V)
- Jumper-selectable LTC line enable/disable and pull-up option - in case anyone needs it
- On-board LEDs for 12V, 5V, POK and DCOK state indication
The only difference from the original is the remote on/off mechanism. The original would have a switch connected to the remote on/off input for enable/disable functionality. The H7864-C needs a positive voltage to (remotely) disable the supply.
Otherwise the H7864-C exceeds its predecessors in total load capability, and in expected lifetime - much better than a used 30-year-old unit anyway.
Tested and working now in a couple of BA23's for several weeks. No issues at all - as expected.
I would still need someone with a PDP11 verify that the LTC signal performs as expected. But that would mean someone with a PDP11 first needs to buy one.
Finally, some observations I can provide, for those of you who may wish to undertake a similar project for a DEC power supply:
1. It is a lot more work than you think
2. Prepare for some serious and tedious re[verse]-[re]engineering work
3. It is more expensive than you think
By the time you have got yourself a set-up, all the required parts, metalwork, connectors, CAD software, prototype PCBs, crimping tools, soldering equipment, pins and what have you, you would have spent a lot of money.
4. It may well end up being less than perfect. One issue I have is the fan connectors - the old DEC fan connector seems to have its two female sockets closer to its surface than the presumably identical connector you can get today. It means the new connector barely holds to the fan connector.
Some gentle filing is necessary, to shave off enough PVC (or whatever) to make the new connector fit nicely on a fan and hold its position.
I will include several photos of a finished unit, inside and out, in the next post. I will also post the PCB schematics and some oscilloscope snapshots showing tests of the PIC microcontrollers' DCOK and POK response to a power failure (in the mixed analog-digital option). I did these tests in order to verify that the timings are within the required 4ms limit. If you do not run a mission-critical application, you don't really need this option.
Now about purchasing one - yes I can build one for you. Please contact me directly. Currently all existing PCBs are built for digital option - as I think no one really needs anything else.
The H7864-C is a fully functional, commercial-grade product but it is intended for hobbyist use. It will probably pass UL and FCC certification without a problem, but this is something I will leave for when an order for 100 units comes along. Which will never happen.
-Alon.
Last edited: