TL;DR: made a new quick-and-dirty PSU for old dbx F900 frames. It works (with some caveats) and I learned a lot from my mistakes.
Big boring post incoming. Above all, I am making this thread to document the things I’ve learned about the very little documented dbx F900 frame, the original early 1980s version with the neon pilot light power switch.
Before you get to into it — this project is NOT compatible with the F900A 9-space frame from the 1990s, only the original with the neon bulb power switch. That frame has a totally different edge connector for the PSU module and is also a 317/337 linear scheme which can be beefed up.
---
Alright story time, I picked up one of the old style dbx 900 8-space racks from a friend for very cheap. Date codes of components in the power supply show it was probably manufactured around 1983. Hadn’t been powered on in 20 years, so I pulled the cards to measure the PSU.
ORIGINAL PSU BACKGROUND:
The stock PSU card lacks a schematic - just doesn't exist anywhere. It uses a metal can 723 scheme for voltage regulation, along with 3055 and 2955 for current boosting, similar to Power One linear modules. It was designed to provide +/-24V unregulated and +/-15V regulated. Mine wasn’t regulating properly: the +15V rail was dead, and the -15V rail was significantly off. Without a schematic, troubleshooting was impractical, so I decided to take a different approach and design something new from the ground up.
The stock PSU is, by design, pretty underpowered and 40 years old. While dbx 900 modules draw 40mA to 100mA each on the +/-15V rails (except for the rare 906 Flanger, which is hungrier), Aphex 9000 modules draw SIGNIFICANTLY more (e.g., the Dominator pulls 330mA on each 15v rail). Aphex 9000 racks use external PSUs with huge Power One modules, handling higher currents easily so they can be fully loaded with Aphex units.
The stock dbx supply is rated at only 1A on the +/-15V rails, and even the later frame with a toroidal PT and LM317/337 design apparently struggles with many Aphex modules. This earlier F900 frame has a Jones Cinch connector on the back for direct voltage rail taps. The design presumably provided this connector in order to power more modules in an unpowered rack (the “F900U”, which I’ve never seen), but the newer 1990s F900A frame lacks the Jones connector.
NEW PSU DESIGN CONCEPTS & OBSERVATIONS:
On a Gearspace thread, some users suggested removing the stock PSU card and just using an external PSU wired up to the Jones connector. However, with design help from forum legend abbey road d enfer, I replaced the old PSU with two 24V SMPS with a CLC filtering scheme into a standard 317/337 regulator setup. Instead of an external box, I laid out a new PCB to retrofit it in place of the old stock PSU card.
I used an LM350 and LT1033 for (theoretically) 3A instead of 1.5A. The PSU card connects via a 10-pin EDAC 306 connector, with joined grounds and 15V rails on thick traces, while 24V traces are single pins. The grounds and their interactions with the Jones connector were sort of puzzling, but attached is a labeled diagram after tracing it out.
Chassis Ground and the 0V meet on the old power supply PCB — the main filter caps go to both traces on the bottom layer of the PCB.
Confusingly, there’s a 100R resistor between the top trace of CGND and bottom trace of CGND — not sure why? Considering both the top and bottom of the edge connector on the rack back plane meet…
I added a jumper on my PSU card for 0V and CGND instead of the PCB, with a 100n C0G cap in parallel with a 10R 3W resistor. Gutting the old transformer and re-wiring the AC receptacle and transformer was the probably tedious part. I've attached a redrawn schematic of how the old PT was wired up with the VAC receptacle / IEC inlet ("dbx F900 - VAC Receptacle Original _ Stock Wiring Scheme.pdf") in addition to a redrawn scheme omitting the PT for my particular project. I left the original neon bulb power switch and used a 220R resistor and JST XH connector for the bulb.
After gutting the old PSU card, I retrofitted my new components onto the existing metal tray/sled. I used Phoenix connectors (despite the Meanwell supplies using JST VH headers). I did this for the sake of “prototyping,” but better idea would have been to just solder all wires directly to my PCB. The new PCB physically fit just fine, aligning with the existing mounting holes, but I drilled extras for additional support. I drilled new holes for the SMPSs and made JST VH harnesses to wire them up.
The original PSU card had extra mounting bolts on the heatsink, but in a studio rack where it isn’t gonna move around a lot, it's physically stable. I used double-sided foam tape on the front panel for additional “support,” though it likely wasn't necessary.
The new voltages were accurate, and I found that my 500 series test extender card/harness from CAPI works with the 900 series modules due to using the same edge connector.
IN USE / PRACTICALITY:
So, does it work? Yes it does. Would I do this again? No, probably not lol.
The LM350 scheme just gets waaaay too hot when trying to run current hungry Aphex modules. Loading the rack with dbx modules is fine, but considering some Aphex modules pull 6x the current of the average dbx unit, things can get overloaded pretty quick and I found that the LM350 gets too hot shuts down. I currently have the largest TO-220 heatsink I can physically fit on the thing (not pictured) and that has seemed to help quite a bit. With my present layout, there isn't really a way to get the LM350 on any sort of larger heatsink.
I have used the F900 frame with the new PSU for 8 to 10 hour tracking and mixing sessions almost every day since completing the project last month (April 2024) and the rig runs totally fine when fully loaded with only dbx modules (and one Aphex 9251 Exciter). No issues, no noise, rock solid.
If I were to do it all over again, I’d just build an external box or a 1U enclosure to go above/below the rack frame with four separate SMPS modules and respective CLC filter scheme (two for the 24v, two for 15v) and run that into the frame's Jones connector, but the OVERARCHING GOAL OF THIS PROJECT was to keep the power supply self-contained within the original frame!
I might try to do a redesign where I will try to fit 4x Mean Well PCB mount modules onto a single PCB similar in size to the original dbx PSU card, two for 15v and two for 24v. The 24v ones can be small because even loading it with 8x of the absolute most current hungry dbx modules still never exceed 800mA ish draw max.
TL;DR FINAL THOUGHTS:
Overall, this setup should be adequate for powering all dbx modules, but it struggles when you throw in very current-hungry Aphex modules. Realistically, the original power supply probably would have crapped out with Aphex modules too, so I guess if anything I accomplished my main goal of bringing it back to life in order to power the modules it was designed to power…
An aside: D'Addario makes excellent line of "Modular Snake" DB25 products, including a connector-free / pigtails-only DB25 breakout cable (model # PW-CFB-01) which retails for only $19.99. This was perfect and a no-brainer for the I/O.
Big boring post incoming. Above all, I am making this thread to document the things I’ve learned about the very little documented dbx F900 frame, the original early 1980s version with the neon pilot light power switch.
Before you get to into it — this project is NOT compatible with the F900A 9-space frame from the 1990s, only the original with the neon bulb power switch. That frame has a totally different edge connector for the PSU module and is also a 317/337 linear scheme which can be beefed up.
---
Alright story time, I picked up one of the old style dbx 900 8-space racks from a friend for very cheap. Date codes of components in the power supply show it was probably manufactured around 1983. Hadn’t been powered on in 20 years, so I pulled the cards to measure the PSU.
ORIGINAL PSU BACKGROUND:
The stock PSU card lacks a schematic - just doesn't exist anywhere. It uses a metal can 723 scheme for voltage regulation, along with 3055 and 2955 for current boosting, similar to Power One linear modules. It was designed to provide +/-24V unregulated and +/-15V regulated. Mine wasn’t regulating properly: the +15V rail was dead, and the -15V rail was significantly off. Without a schematic, troubleshooting was impractical, so I decided to take a different approach and design something new from the ground up.
The stock PSU is, by design, pretty underpowered and 40 years old. While dbx 900 modules draw 40mA to 100mA each on the +/-15V rails (except for the rare 906 Flanger, which is hungrier), Aphex 9000 modules draw SIGNIFICANTLY more (e.g., the Dominator pulls 330mA on each 15v rail). Aphex 9000 racks use external PSUs with huge Power One modules, handling higher currents easily so they can be fully loaded with Aphex units.
The stock dbx supply is rated at only 1A on the +/-15V rails, and even the later frame with a toroidal PT and LM317/337 design apparently struggles with many Aphex modules. This earlier F900 frame has a Jones Cinch connector on the back for direct voltage rail taps. The design presumably provided this connector in order to power more modules in an unpowered rack (the “F900U”, which I’ve never seen), but the newer 1990s F900A frame lacks the Jones connector.
NEW PSU DESIGN CONCEPTS & OBSERVATIONS:
On a Gearspace thread, some users suggested removing the stock PSU card and just using an external PSU wired up to the Jones connector. However, with design help from forum legend abbey road d enfer, I replaced the old PSU with two 24V SMPS with a CLC filtering scheme into a standard 317/337 regulator setup. Instead of an external box, I laid out a new PCB to retrofit it in place of the old stock PSU card.
I used an LM350 and LT1033 for (theoretically) 3A instead of 1.5A. The PSU card connects via a 10-pin EDAC 306 connector, with joined grounds and 15V rails on thick traces, while 24V traces are single pins. The grounds and their interactions with the Jones connector were sort of puzzling, but attached is a labeled diagram after tracing it out.
Chassis Ground and the 0V meet on the old power supply PCB — the main filter caps go to both traces on the bottom layer of the PCB.
Confusingly, there’s a 100R resistor between the top trace of CGND and bottom trace of CGND — not sure why? Considering both the top and bottom of the edge connector on the rack back plane meet…
I added a jumper on my PSU card for 0V and CGND instead of the PCB, with a 100n C0G cap in parallel with a 10R 3W resistor. Gutting the old transformer and re-wiring the AC receptacle and transformer was the probably tedious part. I've attached a redrawn schematic of how the old PT was wired up with the VAC receptacle / IEC inlet ("dbx F900 - VAC Receptacle Original _ Stock Wiring Scheme.pdf") in addition to a redrawn scheme omitting the PT for my particular project. I left the original neon bulb power switch and used a 220R resistor and JST XH connector for the bulb.
After gutting the old PSU card, I retrofitted my new components onto the existing metal tray/sled. I used Phoenix connectors (despite the Meanwell supplies using JST VH headers). I did this for the sake of “prototyping,” but better idea would have been to just solder all wires directly to my PCB. The new PCB physically fit just fine, aligning with the existing mounting holes, but I drilled extras for additional support. I drilled new holes for the SMPSs and made JST VH harnesses to wire them up.
The original PSU card had extra mounting bolts on the heatsink, but in a studio rack where it isn’t gonna move around a lot, it's physically stable. I used double-sided foam tape on the front panel for additional “support,” though it likely wasn't necessary.
The new voltages were accurate, and I found that my 500 series test extender card/harness from CAPI works with the 900 series modules due to using the same edge connector.
IN USE / PRACTICALITY:
So, does it work? Yes it does. Would I do this again? No, probably not lol.
The LM350 scheme just gets waaaay too hot when trying to run current hungry Aphex modules. Loading the rack with dbx modules is fine, but considering some Aphex modules pull 6x the current of the average dbx unit, things can get overloaded pretty quick and I found that the LM350 gets too hot shuts down. I currently have the largest TO-220 heatsink I can physically fit on the thing (not pictured) and that has seemed to help quite a bit. With my present layout, there isn't really a way to get the LM350 on any sort of larger heatsink.
I have used the F900 frame with the new PSU for 8 to 10 hour tracking and mixing sessions almost every day since completing the project last month (April 2024) and the rig runs totally fine when fully loaded with only dbx modules (and one Aphex 9251 Exciter). No issues, no noise, rock solid.
If I were to do it all over again, I’d just build an external box or a 1U enclosure to go above/below the rack frame with four separate SMPS modules and respective CLC filter scheme (two for the 24v, two for 15v) and run that into the frame's Jones connector, but the OVERARCHING GOAL OF THIS PROJECT was to keep the power supply self-contained within the original frame!
I might try to do a redesign where I will try to fit 4x Mean Well PCB mount modules onto a single PCB similar in size to the original dbx PSU card, two for 15v and two for 24v. The 24v ones can be small because even loading it with 8x of the absolute most current hungry dbx modules still never exceed 800mA ish draw max.
TL;DR FINAL THOUGHTS:
Overall, this setup should be adequate for powering all dbx modules, but it struggles when you throw in very current-hungry Aphex modules. Realistically, the original power supply probably would have crapped out with Aphex modules too, so I guess if anything I accomplished my main goal of bringing it back to life in order to power the modules it was designed to power…
An aside: D'Addario makes excellent line of "Modular Snake" DB25 products, including a connector-free / pigtails-only DB25 breakout cable (model # PW-CFB-01) which retails for only $19.99. This was perfect and a no-brainer for the I/O.
Attachments
Last edited:
