3. No risk of "magic blue smoke".
HS are foremost for heath stabily. It works fine without (I ran mine a long time this way) ... just that JFETs are very temp sensitive and offset is somewhat wandering without HS . But not much (provided there is not too much wind inside your chassis
) ... and considering that the volume control is between UGS and output ... neglectible ... I wouln't worry about.
4. BLZ reg will works but you will get even better results with better reg ...
Steen has made a lot of tests on this matter ...
Manu
HS are foremost for heath stabily. It works fine without (I ran mine a long time this way) ... just that JFETs are very temp sensitive and offset is somewhat wandering without HS . But not much (provided there is not too much wind inside your chassis
) ... and considering that the volume control is between UGS and output ... neglectible ... I wouln't worry about.4. BLZ reg will works but you will get even better results with better reg ...
Steen has made a lot of tests on this matter ...
Manu
Re: Practical questions for UGS design
You'd better match them both on Vgs and transconductance.
YMMV, why would you want a consensus ? But as Manu cleverly stated, the output offset - if any - will be decreased by an output attenuator.
If some ugs builders have performed tests on this issue, please report.
go without. No blue screen ahead.
It's an offset stability issue, and it'll help in your case where you plan to use single Jfets.
And it's a matter of long term duration, since the output transistors are hot, and a heat sink can't hurt for them.
If you plan to go the K120/J74 route, use classical TO92 BJTs (BC550, 2SC2240) instead of Zetex's to match the JFets form factor. That will ease the heatsink clamping.
Cheers
Right. Offset, CMRR, gain and susy should be affected.sbear said:
You'd better match them both on Vgs and transconductance.
Please experiment....
YMMV, why would you want a consensus ? But as Manu cleverly stated, the output offset - if any - will be decreased by an output attenuator.
If some ugs builders have performed tests on this issue, please report.
Yes you can
go without. No blue screen ahead.It's an offset stability issue, and it'll help in your case where you plan to use single Jfets.
And it's a matter of long term duration, since the output transistors are hot, and a heat sink can't hurt for them.
If you plan to go the K120/J74 route, use classical TO92 BJTs (BC550, 2SC2240) instead of Zetex's to match the JFets form factor. That will ease the heatsink clamping.
The "original" ugs reg is a pearl type. Again, you're welcome to experiment. But the price of the original reg is low, and it's very simple. You can begin with it, and try other ones once you're used to its "sound"
Cheers
Manu said:
Thanks for the quick response Manu, but yes I have calibrated the ISP board and Pony prog says this is OK.
It has taken some time for me to check it all again as I have had PSU trouble. I am using a 24V transformer to supply the logic and relay circuits. The L7809 gave up the ghost - it was having to deal with 39V input
Anyway I have solved this problem. But I am still getting "device unknown or not present" when I try and read what's on the Atmega64.
chalkandtalk said:Still getting "device missing or unknown device" from pony prog.
- Check the µC solders with a magnifier glass. Any suspect joint ?
- Check if the µC is correctly supplied : 5V on pins 1,21, 52,64
- Check if the HC244 on the prog board is fed : ~5V on pin 20, GND on pin 10, when the µC board and prog are connected via the 10 wire cable (no need to connect the prog to PC, but the µC board must be supplied). Don't plug the display either.
- Any close-up photograph of your µC assembly ?
- Picture of your prog board ?
Cheers
Here is a pic of my prog board.
An externally hosted image should be here but it was not working when we last tested it.
Here is a view of my ATmeg 64 chip.
An externally hosted image should be here but it was not working when we last tested it.
Another view of the ATmeg.
An externally hosted image should be here but it was not working when we last tested it.
I have had some strange behaviour from my PSU. At first I thought it was because 40V was being supplied to the L7809. So I took this out and replaced it with a L7824. I then put a L7809 in between the L7824 and the uC board in the logic power line. Sometimes I have heard the 9V relay on the uC board click and I have measured 5V comming out of the L7805 and also measured 5V at a couple of points on the uC board. However, I now think there wasn't a fault on in the PSU. I powered the uC again and found that the 9V relay did not switch on and also noticed that the L7824 was shutting down, the transformer also started buzzing. I disconnected the the uC from the PSU and turned on the PSU again - no problems, all voltages fine. I then noticed that the ground and out pins of the L7805 are shorted somewhere. I then checked pins 1, 21, 52 and 64 - which are all connected, but they are also shorted to 0V on the L7805. This is surprising because the uC board did power up on occasions. I even used a 9V battery and the relay would switch on and off. I have looked very closely with a magnifying glass and can not see any obvious short.
I wonder if the BC550 could be at fault? As I previosly wired its base to the emitter of the MJE instead of the base of the MJE to the emitter of the MJE
Your advice is greatly appreciated.
cheers
Aidan
PS I will try and get more focussed pis of the ATmeg64.
I wonder if the BC550 could be at fault? As I previosly wired its base to the emitter of the MJE instead of the base of the MJE to the emitter of the MJE
Your advice is greatly appreciated.
cheers
Aidan
PS I will try and get more focussed pis of the ATmeg64.
A better view of the ATmeg64.
An externally hosted image should be here but it was not working when we last tested it.
Thanks for pics, Aidan.
I do not mean to be rude, but your soldering could be cleaner, particularly on the 100K resistors side of the µC : Start by removing excess solder on the µC joints with braid (take your time, don't overheat the chip) and rework all the solders adding very few fresh solder. Then clean the flux (isopropyl alcohol or aceton). Rework in the same way all the solders that seem a bit "ball-shaped" or rounded.
Take care of the small vertical shift of the µC (the vertical rows of the pins ar misaligned with the pads) when reflowing the solders.
On the prog, I see a 330R resistor missing and replaced with another value (near the "to subd casing" label)
Still on the prog, are you sure of the value of the 100pF caps ? They look like bigger values, but I may be wrong...
The first thing then is to sort out the strange supply behaviour. There seem to be a short circuit somewhere... : The trafos hum, no 5V, etc...
Are your 78xx regs on the small supply board heatsunk ?
De-solder the µC board 7805 : remove the screw, gently heat all the three pins and pull the 7805 from the board. Then bring some fresh solder to the pads and carefully clean the holes with a desoldering pump and braid.
Start by checking all the 5V connections : there should be continuity between all the 5V pins of the logic ICs (pin 16 for the 16 pins, pin 20 for the 20pins) and the 7805 output pad. One side of the 100K close to the µC should also be 5V. Use the schematic to find all the 5V connections, and chek if there's a short to ground.
The 9V relay should not be heard, as it's driven by the µC.
The bad BC550 wiring should not be harmful.
As long as you don't find clean 5V on the board where it should be, programming won't be possible... So it's up to you...
Keep us informed of your progress/findings.
Cheers
I do not mean to be rude, but your soldering could be cleaner, particularly on the 100K resistors side of the µC : Start by removing excess solder on the µC joints with braid (take your time, don't overheat the chip) and rework all the solders adding very few fresh solder. Then clean the flux (isopropyl alcohol or aceton). Rework in the same way all the solders that seem a bit "ball-shaped" or rounded.
Take care of the small vertical shift of the µC (the vertical rows of the pins ar misaligned with the pads) when reflowing the solders.
On the prog, I see a 330R resistor missing and replaced with another value (near the "to subd casing" label)
Still on the prog, are you sure of the value of the 100pF caps ? They look like bigger values, but I may be wrong...
chalkandtalk said:
The first thing then is to sort out the strange supply behaviour. There seem to be a short circuit somewhere... : The trafos hum, no 5V, etc...
Are your 78xx regs on the small supply board heatsunk ?
De-solder the µC board 7805 : remove the screw, gently heat all the three pins and pull the 7805 from the board. Then bring some fresh solder to the pads and carefully clean the holes with a desoldering pump and braid.
Start by checking all the 5V connections : there should be continuity between all the 5V pins of the logic ICs (pin 16 for the 16 pins, pin 20 for the 20pins) and the 7805 output pad. One side of the 100K close to the µC should also be 5V. Use the schematic to find all the 5V connections, and chek if there's a short to ground.
The 9V relay should not be heard, as it's driven by the µC.
The bad BC550 wiring should not be harmful.
As long as you don't find clean 5V on the board where it should be, programming won't be possible... So it's up to you...
Keep us informed of your progress/findings.
Cheers
Jim: The resistors are soldered onto the underside of this board, but thanks for taking the time to look. The joints underneath are fine. Even so I have remelted all joints underneath this board.
Cheff:
The missing 330R resistor is really 330R - the colours have just been warped in the digital image. It really has two orange stripes and not two brown ones. I have removed the L7805 and still get continuity between 5V and gnd there. The L7824s and L7809 have heatsinks and when I run the PSU without connecting it to the uC board they produce the right voltages and are quite happy getting slightly warm.
At one point the uC board must not have had this short. I heard the relay switch (although it seems that it shouldn't) on and off and I did measure 5V between out and gnd on the L7805. There is continuity between all the 5V points and the out on the L7805. It seems to me that only the ATmega64 chip itself could cause this changing behaviour, if it's behaving badly. I did read somewhere that some people ruined their ATmega64 chips by using a DMM (with 9V testing) on them. Does anyone know if a faulty ATmega64 chip can short 5V to gnd? I am beginning to suspect that this chip is now dead.
Is it true that the ATmaga128 is more tolerant?
Sorry to be such a pest, but I really would like to get this preamp going.
Aidan
Cheff:
The missing 330R resistor is really 330R - the colours have just been warped in the digital image. It really has two orange stripes and not two brown ones. I have removed the L7805 and still get continuity between 5V and gnd there. The L7824s and L7809 have heatsinks and when I run the PSU without connecting it to the uC board they produce the right voltages and are quite happy getting slightly warm.
At one point the uC board must not have had this short. I heard the relay switch (although it seems that it shouldn't) on and off and I did measure 5V between out and gnd on the L7805. There is continuity between all the 5V points and the out on the L7805. It seems to me that only the ATmega64 chip itself could cause this changing behaviour, if it's behaving badly. I did read somewhere that some people ruined their ATmega64 chips by using a DMM (with 9V testing) on them. Does anyone know if a faulty ATmega64 chip can short 5V to gnd? I am beginning to suspect that this chip is now dead.
Is it true that the ATmaga128 is more tolerant?
Sorry to be such a pest, but I really would like to get this preamp going.
Aidan
Before declaring the µC guilty, have you refreshed, cleaned and checked all the joints ?chalkandtalk said:
If you feel adventurous, try to remove as much solder as you can from the four Vcc pins of the µC with braid, then using a small screwdriver tip and your iron, gently and **very** carefully lift these pins from the pads to interrupt connection. Just a few hundredth of µm are enough, don't fold the pins too much.
At this point, the Vcc pins should not be connected anymore, and re-check the 5V short to ground.
Be aware that there's a big ground "plane" underneath the µC, so that the little shift of the µC that you've got can make a short with other pins. Check this also.
Cheers
