20v to make 10v is a bit hot, but to make 5v is really quite excessive. Realistically 21v is being seen, with 25v capacitors. It's getting a little close. The caps used are 25v and 35v rated, which kinda speaks of a wider separation than 20v-21v which is close enough to call error. Especially if it's is an old analogue meter, as using whole units suggests.
I wouldn't carry on personally. Now it works, you can probably meter out which caps are on which supplies, and which regs are on which secondaries. Loosing a few volts seems like a no brainer. It would be a shame if a 5v reg failed, and let 20v though.
I wouldn't carry on personally. Now it works, you can probably meter out which caps are on which supplies, and which regs are on which secondaries. Loosing a few volts seems like a no brainer. It would be a shame if a 5v reg failed, and let 20v though.
The original transformer is a 10VA type so currents will be low I guess. Knowing the current by measuring the series resistors would tell how to improve the situation if desired.
Coincidentally I have designed a PCB for the aforementioned custom transformer with regulated +/- 15V, +5V with an LDO regulator and +9...12V for a similar device with acceptable losses. A PCB with just the 4 AC voltages I have too. For a second custom 7VA type with 4 separate 4..6V windings offered by that seller I also designed a PCB. If it helps you you can have one for shipping costs. Wired for 230V mains voltage.
Coincidentally I have designed a PCB for the aforementioned custom transformer with regulated +/- 15V, +5V with an LDO regulator and +9...12V for a similar device with acceptable losses. A PCB with just the 4 AC voltages I have too. For a second custom 7VA type with 4 separate 4..6V windings offered by that seller I also designed a PCB. If it helps you you can have one for shipping costs. Wired for 230V mains voltage.
Attachments
Last edited:
The 10VA AC version is 65 x 90 mm. The 7VA AC version is 57.5 x 90 mm. For the transformers themselves and their voltages please check the provided link.
DC versions are not produced yet. I implemented a stupid but popular feature by using an (obsolete) dinosaur symmetrical low noise regulator that I still stock in numbers. Idea is to change to another regulator. Or just do the 2SK170/2SJ74 trick? 😉
An AC version for the standard Talema 700xxK (10VA) with for instance 2 x 7V is also ready. It measures 57.5 x 99 mm.
All have plated through M3 mounting holes, sturdy PCB tracks, 5.08 mm pitch PCB connectors and an onboard TR5 fuse. Pick one if it helps you and pay the shipping.
DC versions are not produced yet. I implemented a stupid but popular feature by using an (obsolete) dinosaur symmetrical low noise regulator that I still stock in numbers. Idea is to change to another regulator. Or just do the 2SK170/2SJ74 trick? 😉
An AC version for the standard Talema 700xxK (10VA) with for instance 2 x 7V is also ready. It measures 57.5 x 99 mm.
All have plated through M3 mounting holes, sturdy PCB tracks, 5.08 mm pitch PCB connectors and an onboard TR5 fuse. Pick one if it helps you and pay the shipping.
Attachments
Last edited:
Sorry was away on the weekend.
Yes A/C. But those dimensions are ok to fit one transformer and I will need to fit 2 in there to provide all the outputs needed.
I don’t want to mess around to much with the circuit.
I will see if I can find a similar transformer with lower output voltage (like 12v or 9v)
They have 4 AC voltages each. That is the specialty.
Anyway, the proper thing to do is to have a schematic with how the PSU is designed.
Anyway, the proper thing to do is to have a schematic with how the PSU is designed.
Last edited:
If right now, this is using a pair of 15-0-15, then is there any need to be a pair? Would this be just as well on a single 15-0-15 as they share ground anyway. It's a shame we still can't see the board, to know where things actually live.
I'm leading toward the idea a 9-0-9 or just a touch more would do everything. These form factor transformers come as a pair of 9s or 10s to make such a thing. Leaving the board asking for a scalpel and links. Bits unseen from above.
It really comes down the the purpose of having two secondaries. Which I suspect is to give two different voltages. However, we have proved just 15v will work for a day.
Does this track, or can someone see smoke signals ?
I'm leading toward the idea a 9-0-9 or just a touch more would do everything. These form factor transformers come as a pair of 9s or 10s to make such a thing. Leaving the board asking for a scalpel and links. Bits unseen from above.
It really comes down the the purpose of having two secondaries. Which I suspect is to give two different voltages. However, we have proved just 15v will work for a day.
Does this track, or can someone see smoke signals ?
How about 7905? Is it just warm like the 7805? Looks like a bigger heat sink might be beneficial.
Wow. An even bigger heatsink would mean they dissipate too much. They dissipate too much as their input voltage is too high. They don't feel very hot as they already have a large area heatsink. When current would be known one could calculate how much power is lost to heat. In a case with too high input voltage one often can see that the power burned to useless heat is higher that the power the load asks. Multiply this by the number of devices and one sees what energy waste is. By choice. Without a valid reason.
78xx/79xx need 2...3V higher input voltage plus the ripple voltage to work perfectly fine. Good designing is to keep dissipation to acceptable levels. Even with let's say a rectified and filtered steady 9V DC a 78xx/78xx would already dissipate (9 - 5) x 0.25A (example) = 1W. Now we have (20 - 5) x 0.25 which is 3.75W to heat. The load is still the same with 0.25 x 5 =1.25W. No device, no power bill, no user, no environment benefits from energy to useless heat conversion.
So to understand and improve stuff one could draw how stuff is done and probably see that regulators are used in series. Then one could use a lower voltage transformer to feed that second regulator. Whether one solves the riddle with 4 transformers with each 1 voltage, 2 transformer with each 2 voltages or with 1 transformer with 4 AC voltages, it comes down to adapting input voltages to each regulator circuit. Just some hints.
Just don't be surprised that even known brands choose such methods. Standard transformers with 1 or 2 windings are cheapest. Those with 2 windings have equal windings. Often a device needs various voltages. Bummer.
It often helps to exxagerate to understand. Calculate with rectified and filtered steady 120V and a virtual 5V regulator with a load of just 1 red LED (10 mA). How much does the LED dissipate and how much does the regulator dissipate?
78xx/79xx need 2...3V higher input voltage plus the ripple voltage to work perfectly fine. Good designing is to keep dissipation to acceptable levels. Even with let's say a rectified and filtered steady 9V DC a 78xx/78xx would already dissipate (9 - 5) x 0.25A (example) = 1W. Now we have (20 - 5) x 0.25 which is 3.75W to heat. The load is still the same with 0.25 x 5 =1.25W. No device, no power bill, no user, no environment benefits from energy to useless heat conversion.
So to understand and improve stuff one could draw how stuff is done and probably see that regulators are used in series. Then one could use a lower voltage transformer to feed that second regulator. Whether one solves the riddle with 4 transformers with each 1 voltage, 2 transformer with each 2 voltages or with 1 transformer with 4 AC voltages, it comes down to adapting input voltages to each regulator circuit. Just some hints.
Just don't be surprised that even known brands choose such methods. Standard transformers with 1 or 2 windings are cheapest. Those with 2 windings have equal windings. Often a device needs various voltages. Bummer.
It often helps to exxagerate to understand. Calculate with rectified and filtered steady 120V and a virtual 5V regulator with a load of just 1 red LED (10 mA). How much does the LED dissipate and how much does the regulator dissipate?
Last edited:
HIFI stuff is often not designed the way a good engineer would do it. I have seen quite some DAC's that run on double 12V AC and regulate down to 5 Volt. Even 15 Volt AC are no problem, just stupid. Often higher and hotter are considered more hiend in HIFI... did I mention stupid?
In this case, to be off the hook, I would measure the current draw of each regulator to estimate the risk of running it on too high load.
.
IMO the extra voltage may reduce the lifespan from maybe 30 years to only 25. Just as an idea. These regulators are build to stand up to 35V input voltage and give 1A output if cooling is in spec.
If you leave such a device on for a couple of hours and touch it with your finger, hot is OK, instant burn is too much. Not 100% scientific correct, but if it is keept at no more than 60°C , what means uncomfortable to touch, it should be quite OK.
To be safe, maybe replace the thermal paste under the regulators. Main cause of such parts failing is dry paste and loose screws.
In this case, to be off the hook, I would measure the current draw of each regulator to estimate the risk of running it on too high load.
.
IMO the extra voltage may reduce the lifespan from maybe 30 years to only 25. Just as an idea. These regulators are build to stand up to 35V input voltage and give 1A output if cooling is in spec.
If you leave such a device on for a couple of hours and touch it with your finger, hot is OK, instant burn is too much. Not 100% scientific correct, but if it is keept at no more than 60°C , what means uncomfortable to touch, it should be quite OK.
To be safe, maybe replace the thermal paste under the regulators. Main cause of such parts failing is dry paste and loose screws.
As far as I have understood, the secondaries of the transformer are OK. Why not measure each resistance? If they are all the same, the transformer had identical voltages on each winding. If not they are different. (Don't dig to deep, sure they could have different power output, but I try to keep it simple)
Also, @jean-paul, they are not made in Germany and optimized for lowest power consumption.
Also, @jean-paul, they are not made in Germany and optimized for lowest power consumption.
After all just as a precaution I ended up installing a 9vx2 and a 12x2 both 20VA same connections as previously
To tell you the truth I had the dac running with the 15v outputs transformers for hours and there was no heat at all. Just a slight warmth on the 7805.
Here is what are the new measurements.
CR2 11.5v on the positive 0 on the negative
CR3 +17 on the positive, -17 on the negative
7905 input -17v
7805 input 11.5v
7810 input 17v
337 input -17v
This is within the 25v of the caps and somehow in the middle of max input for the regulators
The regulators now stay cool
There is also space issue, can only fit 5.5cm wide transformer in there and about 12cm long.
To tell you the truth I had the dac running with the 15v outputs transformers for hours and there was no heat at all. Just a slight warmth on the 7805.
Here is what are the new measurements.
CR2 11.5v on the positive 0 on the negative
CR3 +17 on the positive, -17 on the negative
7905 input -17v
7805 input 11.5v
7810 input 17v
337 input -17v
This is within the 25v of the caps and somehow in the middle of max input for the regulators
The regulators now stay cool
There is also space issue, can only fit 5.5cm wide transformer in there and about 12cm long.
The beauty of higher efficiency design is that transformers can be smaller as well. They don’t need to deliver the extra energy “needed” to be converted to heat. Or: they must be chosen in higher VA rating to deliver undesired extra power for equally undesired heat generation. So in fact higher efficiency design is cheaper in a few ways 😉 The original one was only a quite small 10VA.
Voltages to the 7805 and 7810 are better like this. The 7905 is still too high.
I forgot the output voltage of the LM337 (10V?) but as long as it is at least 3V lower than its input voltage it is OK.
*That still non existing drawing could have been conclusive in finding the optimum.
Voltages to the 7805 and 7810 are better like this. The 7905 is still too high.
I forgot the output voltage of the LM337 (10V?) but as long as it is at least 3V lower than its input voltage it is OK.
*That still non existing drawing could have been conclusive in finding the optimum.
Last edited:
In case someone would like a PCB: I will have these produced for that specific custom wound 10VA transformer. So LM1086-5.0 (or ADJ version) for 3.3 ...5V, 7809...7812 for 9...12V and 7815/7915 (or my favorite the obsolete XR4195) for +/1 15V. Send me a PM if you want one.