@migry
The good news is that there is genuine serial data, the bad news is that it is random for each read. I must be missing something which resets the sequence.
maybe the real programmer has inverted logic? It could explain the random output of the chip. maybe have a look if the data changes on rising or falling edge of the clock signal?
Hi @vvervvurm,
I have actually made some progress with the read out of the device ID "serial bit stream", since my last (big) posting. BTW my programmer is home made and is based around an Arduino Mega 2560 running custom C code.I have DAC control of VCC and Vpp (or Vpg - the high voltage ~12V programming voltage). The circuitry is based around cheap Arduino MCP4725 modules, an OP-amp and a couple of transistors.
So in the past I have made the mistake of continuing to "flog a dead horse", i.e. I have tried to debug a chip which is dead, but which I assumed was working

So I realised that I needed to try some other devices. I had a few at hand. Here is the summary
1) EPM7032LC44-10 : original device, SCOA/B shows serial data but changes from run to run (involves a complete power down).
2) EPM7032LC44-12 : also show random serial data, but is more biases to a string of ones.
3) EPM7032SLC-44-10 (*) : a "newer" device which has JTAG pins - denotes by the "S" in the device code. Seemed to have no activity on SCOA/B.
4) EPM7032VLC-44-12: a 3.3V device (denoted by the "V" in the device code). Bingo. A stable and repeatable bit stream from SCOA and SCOB.

I tried my "read" code and I got stable results but the data looked suspiciously similar to the device ID from address block 0.
4) EPM7032VLC44-12 (**) : also a 3.3V device. Device ID data stream works. Same as the previous "V" chip. The "read" code reads back all address blocks as 80 '1's (all FF == blank).
(*) This device was bought from China. It has laser markings. I did try one using a JTAG programmer and it did program and work.
Just FYI I have bought a number of counterfeit devices (e.g. MC68000, i8088) which originate from a China. The giveaway is that the tops have been sanded down and they have been remarked using laser etching. All my other genuine EPM7032 devices have the text printed in ink. I wanted to see if the JTAG parts could be programmed (or ERASEd) in old-style parallel mode.
(**) This device is black topped. It is a 100% genuine Altera part (I know it's providence). This used to be a trick of the Chinese counterfeiters, until they moved to laser etching. I have a MC68HC000-16 device which came from China via Ebay (yes let the buyer beware). I was able to use IPA to remove the fake ink markings and beneath I found the original Motorola markings of MC68000-8, so not HC and not 16MHz

.
Black topping can be found on genuine devices, but it is of high quality as you might expect. Back in the day (80's/90's?) in the UK RS Components (our Digikey/Mouser) used to sell common ICs such as 7400 LS TTL devices which were black topped and marked with their logo. These were genuine full spec devices.
So success with the 3.3V devices, but not the 5V devices.
I now tried to make sense of the bit stream on SCOA/B, and went back to the IDA Pro disassembly. I found code which manipulated the bit stream. It took some time to figure out what was happening, and I wrote some C code to replicate the machine code in order to play with the bit stream. In one of those "wow" moments a string "ALTURA93" appeared. Close enough

, but after editing the code all that appeared was garbage. After many more hours I found the issue, in that I needed to extend and zero the input array.
For those interested. Each serial out generates 80 bits. in the data area of the code there is an array with 80 elements. It is filled with numbers from 0 to 79 in a logical order (0,5,10,15, ... 70,75, then 1,6,11,16, ... finally 4,9,14, ... 79). It is used to unscramble the 80 bits and create a new string of 80 bits with them simply re-ordered. The "set of 5" appears regularly and must have some relation to the hardware.
I have attached a screen grab showing the captured 16 bits values on SCOA/B and SDOUTA/B (5 * 16 = 80 bits each), and the output from my C code.
I have also attached a grab of the strings embedded in the ALL03 A70X.EXE code. There are 8 strings, which are compared against the descrambled bit stream. The ALTERA93 corresponds to the "V" 3.3V device.