After about 1/2hr operation the image on the dispaly gets scrambled making it unreadable. This problem is mentioned by other users too, but a good solution has not been found yet. So decided to do some troubleshooting myself.
First did a power supply check. In the cold situation Vdd is 3.192V. According the datasheets this should be between 2.85V and 3.15V typical with 3.6V abs max for the display and between 3.0 and 3.6V typical with 4.0V abs max for the PIC. There is only a small overlap of 3.0 to 3.15V for both to be in the typical range. Reduced Vdd to 3.150V to be at the max typical voltage for the display. This is then also OK for the PIC.
The Vdd HF noise at the pins of the PIC is quite large though, I measured >100mVpp with a 100Mhz scope. The PIC datasheet advises good local bypassing with at least 100nF ceramic. At the solder side of the board I decoupled the PIC locally with paralleled 100nF ceramic and 4.7uF tantalum capacitors between pins 13 (+) and 8 (gnd). Also connected pins 8 and 19 directly with a piece of solid wire. This resulted in a reduced HF noise level of 15mVpp (80MHz) on Vdd.
But with both changes done (Vdd decrease and Vdd decoupling), the display got scrambled directly after power up (although less severe, only a few pixels). This triggered me to take a closer look at Vdd. Lowering the voltage seems to worsen the problem. Hence increased the Vdd voltage (in my case to 3.337V). The Vdd is above the typical value for the display however, but below the abs max. This solved the problem, the display does not get scrambled anymore, also after several hours of use.
The main cause seems to be the Vdd for the PIC. Although the datasheet specifies 3.0V as minimum, this appears not to be OK in all cases. A value of close to 3.30V yields better results. The LM317 reference voltage has a slightly negative tempco, and explains the time dependancy. The regulator dissipates about 0.45W and has no heat sink while at the same time the tab is covered by the board at one side hampering the cooling. So it will warm up and the output voltage will decrease a bit. For some cases this might be enough for the PIC to start sending erratic data to the display. Hence adding a small heat sink or maybe just putting the LM317 straight up may already solve the problem. Another way to lower the dissipation is to decrease the input voltage. The unit still works fine at 7.5V supply voltage (although the display back-light will not be as bright anymore, but the effect is limited). Lowering the input voltage can be done by adding a zenerdiode in series with the supply voltage (use a 1W type).
Changing Vdd is possible with either R1 and/or R8 around the LM317. If you choose this option be carefull while doing this and use small adjustments (this is best accomplished by placing resistors parallel to R1/R8, it’s about 10’s of mV changes). Remove the IC’s and the display before the adjustment. Measuring the voltage accross R1 yields the LM317 reference voltage. Using the formula given in the LM317 datasheet allows one to tune Vdd to 3.30V with R1/R8.
For my module I ended up with all changes done: decoupled PIC Vdd, adjusted Vdd to 3.337V, heatsink for the LM317, 2V7 zener. This gives a flawlessly working module.
Unfortunately after building in the module and doing the final checks a scope gnd probe got loose and touched one or more pins of the PIC damaging it. I had to order a complete new kit to get a replacement PIC. Is it possible to obtain a programmed PIC as spare part to complete the now useless second kit?
Thanks for this info, we will look into this.
As a courtesy, we can send a spare PIC, if you supply an address.
Thank you for sending a new PIC, great service which is well appreciated!
For those wanting to use the Audio Analyzer in a headphone amplifier:
Since I have a second module available now, I decided to use it in my headphone amplifier. The headphone has a nominal impedance of 250R and therefore the power scaling is not correct anymore. To correct for this I changed the following components (resistors preferably 1% or 2%):
R4=47k5
R5=R6=33k2
R7=6k82
R9=82k5
R10=1k69 (includes PIC range switch compensation)
R11=5k36 (includes PIC range switch compensation)
R13=481R (includes PIC range switch compensation)
C11=120pF
The power scaling has changed now: the values on the display must be divided by 1000. So milliwatts are actually microwatts and watts are actually milliwatts. As a consequence the 0dBu sensitivity changed from 775mV to 137mV.
The impedance setting on the display will be as follows:
2Ohms is for 63Ohms headphones
4Ohms is for 125Ohms headphones
8Ohms is for 250Ohms headphones
16Ohms is for 500Ohms headphones
For the audio analyzer I have now, the accuarcy is within a few percent.
For normal listening levels only very little power is required.
Thanks again!
Hello,
I am experiencing the same problems. Kit K8098 assembled and when turning on I correctly see the Velleman Logo on the left side and the right side only a few horizontal lines. This output lasts for a couple of seconds and then I get a totally scrambled (flickering) pixel image. Sometimes I can identify something readable like +5 on the top left side.
Operating voltage is around 3.20 V and I tried to decouple the PIC like stated in other posts but that did not help. Has this got something to do with the fact the internal oscillator is used in the PIC rather than the crystal oscillator in the first design? Can I please try a new (more stable) PIC?
Thanks!
hejaled
Most likely, the problem is not PIC-related.
Could be the display.
We recommend to return the kit for inspection/repair.
Please return it to:
Velleman Projects Tech. Dept.
Legen Heirweg 33
9890 Gavere
Belgium.
Please include a detailed fault description.
Hello Velleman Support!
Latest information is this: A friend also bought this kit K8098 because he liked the design and the good pricing and I talked him into to borrowing the dsPIC from his kit and with that one my kit is working fine too!! 
So what to do now? Can I send in my (malfunctioning) PIC for replacement? Or if the price is right I even would want to pay for a new dsPIC for my kit.
Awaiting your reaction,
hejaled
Please return PIC for inspection/replacement.
You can return it to:
Velleman Projects Tech. Dept.
Legen Heirweg 33
9890 Gavere
Belgium
Please include a copy of this conversation.
Hi Velleman Support,
dsPIC is on its way to Belgium.
Thanks in advance!
hejaled
Hi there,
With the replacement received in the mail everything is working fine now!
Many, many thanks!
hejaled