Microelectronics | COMPAQ Prolinea 4/66

Synopsis

My first PC was constructed around a proprietary desktop case and an AT clone mainboard specifically designed for it. Back then, I had friends who had completely generic PCs based on standard AT-class mainboards and normal tower cases of all types -- mostly midi-towers. Also, other friends had completely proprietary PCs made by IBM, Dell, Packard Bell, or Compaq. I was in the middle of both worlds. I always liked the flexibility and modularity of generic PCs, but I also liked the fact that proprietary PCs just worked. People who had them rarely felt the need to update them or switch components and assemblies. In some cases, there was quasi-nothing to update -- think about MCA bus IBM PCs.

In time, I migrated to generic PCs that I assembled all by myself and never looked back. But somewhere in my soul I sought after one of those proprietary 1990s PCs. Over time, I had a few, but somehow I parted with them for one or more reasons. Of them all, I remember mostly the IBM PC 350, the Packard Bell Legend series (which sometimes had a Multi-Media badge instead of the model number), and the Compaq Prolinea 4/33.

While I had both IBM and Packard Bell machines (I wish I hadn't sold them 20 years ago), I never had a Prolinea series PC. Thus, I searched for one of these Compaq branded desktop units for quite some time but failed to find one in good condition. With time, I lowered my expectations and finally found one, albeit in a poor optical condition. The good thing is that the mainboard is in pristine condition, and the power supply is still working. But overall, I think this was the dirtiest PC that I ever bought.

Other than nostalgia, I don't have an immediate use for such a PC. However, I already see it as HP-IB controller on a rack shelve, next to my Hewlett-Packard test instruments. I already have an ISA HP-IB interface that is now installed in another machine. I plan to move it to the Compaq computer. Well, now I have given some weight to this project. In addition, the appearance of this desktop unit relates to the old HP design language.

Without further talk, let's skip directly to the restoration.

Maintenance

Besides a good clean, I would like to preventively replace the mainboard and power supply capacitors. I will also replace the power supply fan with a modern low noise part. Additionally, I'd like to max out the RAM and install a network card. And maybe I will replace the CPU.

Machines of this type are often very simple and easy to work on. Possible issues and unpleasant surprises might be caused by rust or broken plastic parts. This unit suffers from both of them. Good thing is that the two broken plastic tabs were floating around inside, and I was able to glue them together with the front bezel.

Cleaning the Case

As soon as I opened the cardboard box, I was greeted with the most dirty PC that I ever had. Besides the decades of dirt, the case is fairly scratched. On the inside, there are some rust spots. Thankfully, there's no rechargeable battery in this system. Otherwise, it would've leaked for sure, damaging the mainboard and the case. This particular revision of the mainboard has a soldered BR2325 battery, and these don't leak.

Anyway, I dismantled everything and washed the case with soapy water. The plastic parts cleaned up pretty well. But the scratches on the metal case became more apparent. I think that I will have to respray the case.

I will use a corrosion inhibitor on the rust spots inside the case. Until then, I'll leave them like this as they will not extend in a humidity-controlled climate. Somebody had some really greasy fingers and left these ugly stains on the front steel panel. If I ever decide to respray the internals, I will have to clean the metal thoroughly.

I added a personal touch to the status indicators: green for power and orange for HDD activity. Both LEDs were initially matte green. Furthermore, the power LED was dimmer than the HDD activity LED. This indicates a lot of power-on hours.

Then I washed all other mechanical parts and the results were relatively good. Again, with the mention that in the future I might respray some of the inner metal parts.

Let's advance to the next operation.

Mainboard Maintenance

Next, I washed the mainboard with soapy water. After I blew the excess water with compressed air, I let the board in the sun for the rest of the day. It turned out like new.

The PCB is relatively small and effectively crowded with components. Thankfully, there are only two electrolytic capacitors, both in non-critical positions. They are used as part of the audio circuit section. The probability that these parts will ever cause any issues if damaged, is close to zero. But since I have the PCB out, let's replace them anyway.

COMPAQ Prolinea 4/66 Mainboard
IdentifierValueQtyNotesMouser Number
C504, C509100 uF / 35 V2Polymer Capacitor80-A759KS107M1VAAE31

These capacitors are located next to a TDA7052 mono audio amplifier integrated circuit. Inspecting the datasheet reveals that this IC does not require an output coupling capacitor. So I think they might be used for local power supply ripple filtering. But it makes me wonder why a 35 V rating? I don't know what that COMPAQ rebranded chip does, but I'd suspect it might also be part of the audio circuitry, though I cannot find any datasheet. I'm curious whether it's a logic IC or some kind of analog (pre)amplifier circuit. By comparison, there is a LM324 quad op-amp IC that has a tiny package.

Anyway, here's a picture of the replacement capacitors already installed.

Before moving on, let's measure the old capacitors. I suspect they are in good condition.

CAPACITOR DATA
BrandTypeStyleRatingCapacityESR
NichiconElectrolyticRadial100 uF / 35 Vxxx uFxxx Ω
NichiconElectrolyticRadial100 uF / 35 Vxxx uFxxx Ω

The next step is to solve the issue with the battery. A soldered BR2335 lithium carbon-monofluoride battery would've discouraged most users back in the late '90s when these were end-of-life products. Well, it kind of annoys me as well. My initial thought was to bypass the battery and make use of the existing external battery pin header. I think a 3 x AAA battery holder would look a bit out of place. But then I thought about adding a standard CR2032 battery holder. Surprise! I couldn't find any with that 3-terminal disposition. Well, I thought about making my own interposer PCB. But that would take far longer than the time I have allocated for this project. Thus, I ordered a CR2032 and a CR2450 battery with three solder terminals.

Solderable CR2032 and BR2335 batteries have slightly different terminal rasters, but with a bit of care, a CR2032 does fit the mainboard holes. Finally, I installed the larger capacity battery since it was a direct fit. In addition, it should give me a few years of trouble-free operation. In the meantime, I might just craft that interposer PCB.

After the battery replacement episode, I reinstalled the mainboard in the case and inserted the ISA riser card that I had previously washed with soapy water. The empty slot on the right is for a proprietary cache module.

These are hard to come by, and at this point, I am absolutely at peace with the thought of never finding a cache card for this mainboard. Nor any VRAM extension board. Thus, let's move on.

Power Supply Maintenance

The classic AT-style power supply is a 145 W unit. The electrical and mechanical construction is of good quality and the components are mostly from reputable manufacturers. The amount of fine dust present in the PSU reveals that it has been turned on for extended periods of time in a harsh environment. Possibly in a school in the countryside.

First, I washed everything with soapy water. After drying up, I proceeded to inspect the unit. As with any old power supplies, I checked for the dreadful RIFA capacitors. Thankfully, there were none. Most of the electrolytic capacitos are made by Chemi-Con, but there are some smaller ones from obscure brands.

The following list contains the parts that are due for replacement.

COMPAQ Prolinea 4/66 Power Supply
IdentifierValueQtyNotesMouser Number
C204, C209220 uF / 25 V2Polymer Capacitor80-A755KS227M1EAAE25
C30147 uF / 50 V1Electrolytic Capacitor710-860040673002
C305, C601, C6034.7 uF / 50 V3Electrolytic Capacitor710-860160672005
C501, C50247 uF / 25 V2Polymer Capacitor80-A750EK476M1EAAE40
C403, C702, Cxxx10 uF / 50 V3Polymer Capacitor80-A759BQ106M1HAE105
C100 uF / 25 V1Polymer Capacitor80-A755KS107M1EAAE25
C1200 uF / 16 V1Polymer Capacitor647-RNL1C122MDSBSQ
C3300 uF / 10 V2Electrolytic Capacitor661-EKY160ELL332MK3
C3.3 uF / 450 V1Electrolytic Capacitor667-ECA-2WM3R3
C470 uF / 200 V2Electrolytic Capacitor647-LGR2D471MELZ40

I normally select replacement capacitors from the same manufacturer for all of my projects. But for this power supply, I just picked up whatever I quickly found while browsing the Mouser parts catalog. Besides the well known brands, this time I also picked up some Würth Elektronik branded capacitors. I don't know whether they're good or not, but at least they are new. I guess time will tell if I did the right thing.

Let's take a look at the measured parameters of the old capacitors. Again, I suspect they are in good condition.

CAPACITOR DATA
BrandTypeStyleRatingCapacityESR
Chemi-ConElectrolyticRadial100 uF / 35 Vxxx uFxxx Ω
NichiconElectrolyticRadial100 uF / 35 Vxxx uFxxx Ω

Here's the power supply PCB with the new capacitors installed.

The original fan is still in good condition, albeit very dirty and noisy. But the bearings are still OK. I decided to replace it with a 12 V Noctua NF-A8FLX fan. As an HP-IB controller, I'd disconnect the magnetic hard disk drive, and combined with a CF card, this PC would be totally silent.

And here is the completed PSU assembly.

The power supply restoration was straightforward and easy to accomplish. The machine can now be left powered on without any supervision as the risk of something blowing up or catching fire has been greatly reduced.

Future Plans

Since clock-doubled 486 CPUs greatly suffer from the missing L2 cache, I'd like to source a 486DX / 50 MHz CPU, and use that one instead. There might be some stability drawbacks, but I will strive to overcome those. Hopefully, the chipset and the peripherals on this mainboard can cope with the 50 MHz CPU clock.

I am also thinking about a method to restore the case badge. The red paint is chipped, but the substrate is pristine. Maybe it can be reworked with some translucent vitrail paint.

Finally, I am thinking about respraying the outer metal case. And possibly the inner case as well. I'm not sure if I should do it myself, or go to a paint shop.

Curiosities

The internal mainboard documentation actually lists this machine as a Deskpro XE instead of a Prolinea. This leads me to think that these models were pretty similar and were produced in parallel.

The designers went wild with the video section. They used a little known Motorola video controller chip that goes by the name of XC02SH007NS07. In addition, whoever did the PCB layout had a weird notion of symmetry. Take a look at those four pin headers next to the video RAM chips. There is absolutely no logic in having them such wildly positioned. Crafting a DIY VRAM extension board will most likely be delayed by lots of measurements with the precision caliper. Just... crazy.

There's no board real estate issue as the mainboard is a 6-layer PCB. So they had all the space they needed to route the wires. Anyway, my guess is that the tracks are autorouted by software.

Don't get me started on the cache RAM support. They used this weird PCI-like brown slot that accepts a proprietary cache card containing the SRAM chips and the cache controller chip. Well, it is almost like they didn't want this machine to ever be user upgraded. 486 chips, and especially clock-doubled (or -tripled) ones, such as the DX2 or DX4, suffer great performance losses without cache. In my opinion, such a system without cache would better work with a 486DX / 50 MHz.

By soldering the battery on the mainboard, the designers just sent a clear message that these systems are consumables. Good thing they provided an external battery header and associated configuration jumper for backup. At least they didn't use any rechargeable batteries. Anyway, hobbists will always find ways to overcome original design limitations. As I previously said, I can craft an adapter (interposer) PCB with a CR2032 battery holder. But I'm not sure if it's worth it. It's easier to solder a new BR2335 battery instead. Or for this matter, any 3V lithium-based battery with three solder terminals with the required raster. Just like I did with the CR2450.

As a side note, one of the rust spots is clearly visible to the right of the battery.

For standard PC users (and builders), a mainboard with integrated RAM chips might sound outdated. At least, I think about the old IBM XT systems with those large array of onboard RAM chips. For those who have worked with similar computers from Dell, Packard Bell, IBM, and possibly others, having onboard RAM might be as expected. But for me, it's a curiosity. Why did they choose to go that way? It beats me. Probably for cost-cutting purposes so that they could sell PCs with 4 or 8 Mb of RAM without any SIMM. Could be.

However, to fully support my plan, I just hope the onboard 70 ns RAM chips can cope with a 486DX running at 50 MHz. If I were to replace the on-board RAM chips, then I'd have second thoughts about the 50 MHz processor.

Another curiosity would be related to the amplified audio output. That's an interesting approach, but quite uncommon. I imagine you could use the software mixer to control the volume of the integrated sound card. But then again, dedicated speakers with internal amplifiers were a better choice back in the '90s. Nowadays, I just pass everything through a (Sansui AU-8500) power amplifier and I either use my headphones or the big speakers.

Conclusions

To me, this PC is a contrast between quality components, overengineering, careless layout, and updateability (but with unobtanium parts). Don't get me wrong, it has its own charm. But sometimes, it's a pain to get along with it. Especially since it uses that floppy disk-based setup program. Well, you can shift everything to a hidden hard disk drive partition. But that's even weirder.

I guess these kinds of systems were IT infrastructure engineers' dreams. All identically configurable, predictable, easy to deploy, and easy to maintain -- if you know the procedures, of course.

But hey, my dream finally came true and I have a chance to play with one of these Prolinea desktops. And I enjoyed every step of this restoration. Well, maybe the reality check with the cache and VRAM expansion boards set me off for a bit. But then again, it's not like I live in the mid-1990s and this is my sole PC. As I said, I just have to adjust my expectations. This time, by lowering the stakes.

I connected the PC to a free channel on my KVM switch, and I played some old MS-DOS games, including Doom and Wolfenstein 3-D. The look and feel of this old desktop, combined with the floppy and hard disk drive sounds, give that nostalgic '90s feeling.

Up until further upgrades, I will follow my initial plan of converting this machine to an HP-IB controller. But that would be the subject for another essay.

Your Help Matters

This content is provided as-is and is not for commercial purposes. It reflects my experiments and research and should be treated as such. I release my work to the public for educational purposes. I did all this at my own expense and in my free time. So if you like my work, or find it useful or inspiring for your projects, please consider making a donation.


Copyright © 2004- Alexandru Groza
All rights reserved.
VER. 1.0 | REV. A