BRAINSOFT'S MULTIPLE I/O PODULE

Reviewed by Lee Calcraft

The spec of Brainsoft's new podule for the Archimedes is truly impressive. On just one half-width podule they have fitted:


A mono video digitiser
A stereo sound sampler
An analogue port compatible with that on the BBC Micro
A pair of RS 232 ports and three ROM sockets

And all this costs just £107 +VAT.

The board is extremely well-made, and is easily fitted into the machine, assuming that you already have a podule backplane. It has three sockets at the rear, a BNC for video input, a standard 15 pin D connector for audio and analogue input, and a non-Acorn standard 25 pin D connector for the RS 232 ports. The driver software for the podule is held in a 32K ROM on the board itself, and this is automatically installed when the machine is powered up. Unfortunately, although this provides a number of low level SWI calls, only four star commands are provided.

The present version of the Multi-Podule I/O board (version 1.2) as marketed by Brainsoft, does not in fact meet with the full specification above in the following respects:


No disc-based software is provided to run any part of the podule.
The ROM filing system is not implemented, so that ROMs may not be used with this board.
Sound sampling is implemented in mono only.
Only one of the two RS 232 ports is functional.

In the near future Brainsoft are hoping to release a new version of the board which will overcome these shortcomings. Upgrades will be available for a small sum.

THE VIDEO DIGITISER

Two star commands are provided for using the digitiser:
*BVidSamp12 and
*BVidSamp15
Both require a filename, and will grab a frame (in reality an amalgamation of 100 frames - see later), display it on the screen (mode 12 or mode 15 depending on the call) and save it to disc. It is a shame that there is no star command to perform the task without saving to disc - though you can achieve this easily with a pair of SYS (or SWI) calls:
SYS &408C0:SYS &408C2
for mode 12, and:
SYS &408C0:SYS &408C1
for mode 15.

In use I found Brainsoft's podule software to be somewhat unstable. If you call for a frame to be grabbed, and no signal is present, instead of the "No video signal" message which you get from Watford's digitiser, the machine just hangs. You can't even restore it with Ctrl-Break. Ctrl-Reset will get you out, but after this the podule's star commands and SWIs cease to be recognised. Only by switching off the computer can you make things work again. I found that the software hung up on other occasions too. If you tell it to perform a mode 15 frame-grab, but the computer is in mode 12, it will also hang up, and so on.

The picture quality of the grabbed image was also most disappointing. In order to keep the cost of the unit to a minimum, Brainsoft have used a relatively slow D/A (digital to analogue) converter to scan the video image - in fact the same converter is used for the video digitiser, the sound sampler and the analogue port, so they can't be used simultaneously. There is also no RAM on the board (Watford's digitiser has 160K of on-board RAM for fast frame storage).

Because it is a relatively slow device, the A/D is not capable of scanning a single video frame in real time. It actually scans around 100 frames to build up its image. This takes some two seconds or so, and for the very reasonable price of the board, two seconds seems quite acceptable. But there is a problem inherent in this approach which may be insoluble. In all tests which I carried out, both with a VHS-C video camera, and with a good quality video recorder (used in freeze-frame mode to halt the picture for 2 seconds), all images were very noisy. This manifested itself as distortion on all horizontal lines of the picture. In spite of advice from Brainsoft, I was not able to overcome this.

THE ANALOGUE PORT

The Brainsoft analogue port simulates that on the BBC Micro, or on Acorn's Archimedes I/O podule. You can access it using Basic's ADVAL function, so that the board's dearth of star commands is no drawback here. In tests, I found that although the port caused values to be returned by ADVAL, its output was far from stable. A short program to print out the voltage at the analogue input at intervals of 0.1 seconds gave the following sequence of of readings with a stable resistive source:
14592, 15872, 15616, 14336
and at the low end, fluctuations were even greater:
512, 256, 2560, 2560, 512
and so on. The same input applied to the analogue port of a BBC Master gave constant and unchanging values.

SOUND SAMPLER

The podule's sound sampler will digitise audio signals input at the A/D connector. Two star commands are provided for this purpose:
*BSoundSamp
wrongly given in the manual as *SoundSamp, and
*BSoundPlay
confusingly given in the manual as *PlaySound.

When you get the name right, both commands take a filename as parameter, and store a 4 second sample to disc or replay it, respectively. It is a shame that there are no star commands for taking samples of other lengths or for replaying them without saving them to disc. There is also a problem with the RMA. The sample is built up in the RMA, and although the manual doesn't explain this, you need to configure some 250K for this purpose. What is really annoying is that the RMA is not cleared afterwards, so that if you repeat the command *BSoundSamp, it will fail with a No room in RMA message.

If you want to take samples of different lengths, you can use one of the SYS calls provided, though I failed to get these to behave properly. During sampling the only way to escape is to use Ctrl-Reset (Ctrl-Break would not work), and this again caused the computer to stop recognising any of the podule commands. It was as if the Arc had had enough of this podule, and I must admit to feeling the same way about it myself.

As to the quality of the sampled sound itself, this was seriously impaired by a high level of white noise, which was itself distorted.

CONCLUSION

I would like to think that the board reviewed here was faulty, and that there is nothing fundamentally wrong with the product itself. However, by no means all of the shortcomings of this product can be put at the door of a single board fault. The system software is obviously in need of an overhaul, and I have my doubts about the feasibility of scanning a video frame by sampling from 100 frames using a relatively slow A/D device. In such a set-up, each frame must be identical to the last, but more importantly the timing of each individual digital sample (i.e. the timing applied to the sample-and-hold device) will be absolutely crucial if image distortion is to be avoided.

However, Brainsoft's attempt to provide a useful cluster of functions on a single board and at a very reasonable end-user price is laudable. If they can supply us with a new version with these problems fixed, we will report back.
ProductMultiple I/O Podule
SupplierBrainsoft Ltd
 106 Baker Street,
 London W1M 1DF.
 Tel. 01-486 0321
Price£123.05 inc. VAT + £3 p&p