Unilumin Upad 2.6mm led screen repair

I regularly repair tiles of Unilumen 2.6mm Upad that have broken pixels. Since the pixel pitch (distance between the center of the pixels) is only 2.6mm, that leaves only about 0.5mm between the pixels. Not enough for soldering with conventional means. To do this, you do require good eyes or magnification as well as an hot air station.

96×96 pixels on a 25x25cm tile

Example of damage by ripped off pixels as on a tile received by customer. Note the distance between the center of the pixels is only 2.6mm. Pixel dimensions are 2.1×2.1mm.

---

It took me quite a while to become proficient at replacing the pixels and performing repairs on pads that had been ripped off.

To repair damaged pads, one can use special (very thin) paper-like circuit board that you can order from Unilumen.

In the process of learning how to deal with these repairs, I created some documentation to make this somewhat easier. I wish I would have had these drawings when I started out as this would have allowed me to get better results sooner. As such, I think it is only fair to share these drawings hoping to help someone else. Below are some of my notes documenting orientation as well as traces on the board itself as well as on the paper. Pictures of the paper are taken through a stereoscopic microscope.

Layout of the tile traces

Front side of paper with RGB traces

Backside of the paper with GND connections

Hope this helps someone!

Advertisement

HP 8711A: modifications, tweaks, refurbishing

I recently acquired a fairly well preserved HP 8711A network analyser. As sold, this is an economy model, offering only scalar measurements (although vector information can be acquired through the HP-IB bus).

Conversion

What makes this unit interesting, is that the hardware is the same as the never released 8712A, which has vector capabilities. Some changes in the bootrom allow the user to “convert” the unit, making it behave like a full fledged VNA.

Credits for this rom modifcation go to Massimo Porzio (IK1IZA). As far as I know, he is the one who took the time to figure this out. I also found a lot of information on the site of Michal Lewczuk (SP2XDM).

Before doing anything, a backup of the correction constants was made to disk.

Unit boot screen as received:

I used the TL866 USB High Performance Programmer which can be bought online for around €40. Download the software while you wait for the package to arrive, download speed reminded me of the late nineties…

The bootrom with attached label (on the A1 CPU board, removal of the CRT/PSU module is necessary):

Contents of the ROM at address 0x1FFA0 prior to modifications:

After modifications:

New bootscreen (SRL and fault location enabled as well):

Smith chart as proof of vector capabilities:

CRT realignment and brightness setting

Some adjustments on the CRT were done:

• When looking from a normal viewing angle with the unit placed on the desk, the alignment of the options on the right of the screen seems “off” when compared to the location of soft buttons next to them.
• The screen was rather dim.
• Sizing of the display compared to the cutout in the front panel could be enlarged.

I marked the locations of the front panel cutout and the top and bottom buttons on the CRT:

After disassembling the CRT/PSU module and attaching the PSU to the back, followed by some creative cabling, the unit was powered on while leaving access to the alignment potmeters:

The positions of the potmeters before adjustment were marked:

After adjusting the height/width and brightness, the screen looked much better.

Power supply patch

According to a ECN from HP, units with serials between A00000 and 3325A00941 need a 3W, 680Ω resistor between pin 9 and 16 on J5 in the power supply.

Before:

Resistor added and pcb cleaned:

Speaker “modification”

The A models of this network analyser had an issue where noise would get into the audio circuit, causing a high pitched, squeeling noise. When used in a silent environment this was really bothersome, so I first tried replacing the speaker by a version with a slightly different resonant frequency.

This didn’t help enough, so I ended up with this ugly (but fully working) solution:

Cleaning and restoring the unit

A lot of time was spend cleaning and restoring the unit, as to give it as much of its original appearance as possible.

• The inside of the case was cleaned, connectors on the backplane (which are notorious for causing problems with this model) were cleaned using a combination of MEK, IPA, soda, warm water, patience and lots of love.
• Front panel and keypad were cleaned, and the yellowing of the plastic due to age and UV was reversed through the use of high concentration hydrogen peroxide. Dents, holes and cuts were fixed.
• The vinyl was cleaned, restored and treated.
• Internals of the disk drive were cleaned.

Documenting this is worth a separate post.