SA-9800 PCB Layout Info

[TalYWaun]

Move Q11–Q14? If there was somewhere obvious and easy to get to then maybe. If they really have a heat problem then maybe a solid, pcb mounted, extruded heatsink would be my first step. But considering the 120mA indicated on the schematic they shouldn’t be working hard.

I have spaced out Q1 and Q2 and given them significant copper pads top and bottom. The position isn’t ideal but hopefully with the equivalent components and the copper to sink heat into, they should fare better than the original parts.

I like your thinking on the power distribution but had put together a different approach . How about a variable Voltage supply feeding J7?
My problem is finding a regulator that will run on the available line voltages. The 32V at J5&6 is such an easy candidate. But as you point out:

“ Another reason is that the J30 IIRC is mostly for control circuitry and not audio so putting the VFD grid on this keeps the supplies for the amplifier sections away from noise.”

My version (1C) with the emitter follower (that evolved a bit) added, is pretty much done. But, VFD Grid or alternate display options when the VFD finally gives up, an AUX Power Supplies option could be useful.

Using 1C as the base I have made the board 28mm longer and added the PSU unit there.

So Version 1D.

Will Route a couple of slots across the board on the current boundary. Don’t want the mod? Just snap it off and throw it away.

So, minimally invasive of 1C pcb and there is space in the chassis.

New fuse off J30 feeding 2 x DC to DC Converters. The one I have sketched in is a V7824W-500. 52V in, 24V 300mA out. That’s for the Grid Supply. So V Out on J7 is adjustable 7-22V 250mA.

In the space underneath the fuse holder I am going to place a 64pin STM32F4 MPU that wakes up when it gets warm and then wakes up occasionally to monitor the system. It could also drive an alternate display, but let’s start simple.

As the MPU is to be minimally invasive of the 1C board I may add another 2 layers to the PCB to route the tracks to thermocouples that could be added near power components and other A/D devices. Plus my wish for a remote volume control would be step nearer.

 

[QSilver]

I'm just going through this slowly. But I have a couple of points so far.

The snap off addition is a pretty interesting idea. Only thing to be weary of is the wires coming out through the chassis there. On multi-voltage versions, the 5 or 6 extra wires from the voltage selector are around there so I'm not sure if they would all happy tuck underneath a board. I don't want to shoot the idea down, but I just thought I'd point out a potential sticking point.

I may have got the VFD grid and other voltage control a little mixed up so I'll just re-cap. The J7 connection really needs to be between 17-19V because although this is the voltage to illuminate the function indicators in the VFD, it is also the power supply for the IC's on the VFD board which drive the power meters. I recently checked the voltage tolerances on the drive chips on a data sheet and IIRC, they can't take much over 20V, certainly not 30V so this regulator really should just be a clamped 17-19V to avoid causing problems. As mentioned in other threads, the SA-7800 and SA-8800 have a regulator on the VFD board for this but the SA-9800 doesn't. It just makes do with this Zener....

The Grid on the VFD has to come from another source because the voltage needs to be able to get up to a maximum of 63V and I think one reason to drive it from the supply I suggested is because over time the VFD is drawing more current as it ages which causes the voltage to drift and then the tube begins to be over driven. Using a regulator here will clamp the voltage to one specified by the installer and then the voltage will be maintained irregardless of current. My concern is that if the circuit is set to output a maximum of 300mA, you've set a ceiling and the original situation may repeat, however I may be over thinking the current requirement. When I performed this modification on an SA-9800 recently, it did begin to draw a noticeable amount of current, however I forgot to measure how much it was precisely.

Replacing the VFD is something that I am thinking about on and off. But its not a top priority for me at the moment.

Can I ask why you chose to use DC-DC converters?

I've had an idea about saving space on the board. You could take out the two axial 220uF 100V capacitors and replace them with radial capacitors? I've done this once or twice and the Nichicon PW capacitors can fit here provided you pick the correct height. Might not free up much space, but it could help?

 

[TalYWaun]

The Power Transistors connected at 7-10 of the Power Amp board are causing a problem, disconnect them and everything else seems fine, even the VFD is working.

My unit has a UK Voltage selector switch, not seen the US version. I kept the expansion to about 2/3 the space available. I will try and keep access space around the connectors. Or they could be repositioned to the other edge.

Ah, I thought J7 was the grid voltage. So is J13 the grid voltage (going to the bright/dim switch)?

In work we are using an inexpensive little display for a touch screen. 1200 x 780 full colour. It would take some serious engineering just to install, but could be an option.

The DC-DC converter is just a placeholder, sort of. Tried looking for a transformer but couldn’t find anything that fitted. Will do a better search when I have finished sketching out what I want.

Moved the MPU to the bottom of the board under the transformer. From there the tracks run through an (optional) inner layer. I have put voltage sensors on J19-J13 and at R1. SOT23 (little) heat sensors under Q11-Q14, Q1, Q2. Could well be too noisy for home use but will come in handy for testing.

Testing – how? Running it in an environment chamber – no problem. At full load without upsetting everyone in the building could be trickier. Maybe 4 X 600ft of mains wiring for a load?

With the MCU on the main bd it needs a 3V3 supply. Will add one near the MPU which could even run the optional pilot LED. That would free up some space for new supply options that will be part of the 1C build J7 & J30 supply.

1D getting near the end of the sketch phase. Need to move 2 voltage sensing points as they are on negative voltages.

Need to run the pilot LED from a biased transistor for 1C or switched on 1D (LED Flashes for power up problem / pulses faster as it gets hotter). Thinking of adding a 16 channel SPI GPIO chip + connector (expansion option) but only if I can squeeze it between the existing data layer tracks.

Yes, I know there is a very good chance that it could easily be damaged but as long as it holds together in the lab under testing then I would be happy.

So, suggestions welcome.

The data layer is green in the pics.

 

[QSilver]

Ah, I thought J7 was the grid voltage. So is J13 the grid voltage (going to the bright/dim switch)?

Yes, thats correct, but as it is unregulated, its no good for the VFD anyway.

The DC-DC converter is just a placeholder, sort of. Tried looking for a transformer but couldn’t find anything that fitted. Will do a better search when I have finished sketching out what I want.

Looking at J7, a simple regulator will work here. It just provides power to the VFD IC's and a voltage to illuminate the function indicators. It can be run from the same position as it was originally, just with a regulator rather than a Zener. This helps with the VFD problems too.

Running it at full load, you just need two large 8 ohm resistors and then keep them cool.

 

[TalYWaun]

It’s alive!

Heard sounds through both channels. The right channel looks like it needs setting up (or there is still a problem there), but so far so good.

Power Amp board Layout:

 

[TalYWaun]

The new PSU board is almost ready to send to manufacture, just need to finalise the MPU choice and check compatibility.

For now, I’m going to reassemble the amp and take it home for endurance testing over the Christmas period. Damn it’s going to be nice to have my amp back home.

New blue pilot lamp fitted (mouser 696-SSL-LX2573USBD). It’s a 20mA part currently running at about 8-9mA but it’s still too bright for my liking. The really great part – it exactly fits the hole when the original orange light pipe is removed.

So Seasons Greetings to everyone and I’ll be back when the new board is running.