The rating allows for the higher transient conditions during startup. Its the steady state conditions that are more important anyway, since the issue is thermal...
(calculate the heat dissipation as a function of input output voltage differential and you'll see what I mean)
Anyway, the optimal operating conditions for all these regs is to have the input voltage about 4-8 V higher than the output. No reason to exceed this, and no reason to worry about the max rating.
(calculate the heat dissipation as a function of input output voltage differential and you'll see what I mean)
Anyway, the optimal operating conditions for all these regs is to have the input voltage about 4-8 V higher than the output. No reason to exceed this, and no reason to worry about the max rating.
I hadn't read any it as criticism. Your answer was perfectly correct, but I was concerned when I attempted to extrapolate back to what prompted TBs original question.
Andrew said he will use 12V windings, so he'll end up with 17 V rails, which will do just nicely for the 12V LT1085/1033 regs he plans to use... 25V rated caps will also be sufficient. So as far as I can tell theres no reason to delve into the depths of the datasheets...
Andrew said he will use 12V windings, so he'll end up with 17 V rails, which will do just nicely for the 12V LT1085/1033 regs he plans to use... 25V rated caps will also be sufficient. So as far as I can tell theres no reason to delve into the depths of the datasheets...
Hi,
for the sake of an example, let's do some numbers.
In the UK the tolerance on mains supply is +6% -10% around 240Vac i.e. 216Vac minimum to 254Vac maximum.
For small VA transformers the regulation can be in the range 15% to 30%.
For a single bridge rectifier on a dual polarity supply the Vdrop into a near opencircuit (smoothing cap leakage only) can be about 0.5V and when outputting significant current approaching 0.8V.
Using those assumptions we have the following for a 230Vac:15Vac +15Vac transformer.
at 230Vac output =15Vac.
at 254Vac output =16.56Vac.
at 216Vac output = 14.09Vac.
worst case peak voltage on the smoothing caps is 16.56*1.3*1.414-0.5=30Vdc. (goodbye to long life on 25Vdc smoothing caps).
With a light load the worst case maximum PSU voltage will drop about 1.2V from peak to about 28.8Vdc. Dissipation in the regulator passing 100mA is about (28.8-12) *0.1=1.7W (it needs a heatsink).
Now let's look at minimum voltages.
at 216Vac and a heavy load (say 200mA). the minimum PSU voltage will be 14.09*1.414-(0.8+1.2+1V)=16.9Vdc.
From this we need to subtract the ripple voltage on the snoothing caps. let's say this is 1Vpp.
we have 16.9V-1V=15.9Vdc at the bottom of the worst case low voltage on the input to the regulator. Vdrop for the negative regulator is specified as 3V. Subtract this from the lowest input voltage and you have 12.9V. This is >=the required output voltage. There it is OK.
BUT,
15Vac risks damaging both the smoothing caps and the regulator if no heatsink is fitted.
The above is roughly what would result from a 6VA to 12VA 15Vac transformer.
If you use the recommended 12Vac transformer you must ensure the lowest voltage, at the bottom of the ripple exceeds the (output voltage +Vdrop). This is ensured by using a higher VA transformer and more smoothing capacitance but do not use too much to be safe. Excess smoothing brings other avoidable problems.
for the sake of an example, let's do some numbers.
In the UK the tolerance on mains supply is +6% -10% around 240Vac i.e. 216Vac minimum to 254Vac maximum.
For small VA transformers the regulation can be in the range 15% to 30%.
For a single bridge rectifier on a dual polarity supply the Vdrop into a near opencircuit (smoothing cap leakage only) can be about 0.5V and when outputting significant current approaching 0.8V.
Using those assumptions we have the following for a 230Vac:15Vac +15Vac transformer.
at 230Vac output =15Vac.
at 254Vac output =16.56Vac.
at 216Vac output = 14.09Vac.
worst case peak voltage on the smoothing caps is 16.56*1.3*1.414-0.5=30Vdc. (goodbye to long life on 25Vdc smoothing caps).
With a light load the worst case maximum PSU voltage will drop about 1.2V from peak to about 28.8Vdc. Dissipation in the regulator passing 100mA is about (28.8-12) *0.1=1.7W (it needs a heatsink).
Now let's look at minimum voltages.
at 216Vac and a heavy load (say 200mA). the minimum PSU voltage will be 14.09*1.414-(0.8+1.2+1V)=16.9Vdc.
From this we need to subtract the ripple voltage on the snoothing caps. let's say this is 1Vpp.
we have 16.9V-1V=15.9Vdc at the bottom of the worst case low voltage on the input to the regulator. Vdrop for the negative regulator is specified as 3V. Subtract this from the lowest input voltage and you have 12.9V. This is >=the required output voltage. There it is OK.
BUT,
15Vac risks damaging both the smoothing caps and the regulator if no heatsink is fitted.
The above is roughly what would result from a 6VA to 12VA 15Vac transformer.
If you use the recommended 12Vac transformer you must ensure the lowest voltage, at the bottom of the ripple exceeds the (output voltage +Vdrop). This is ensured by using a higher VA transformer and more smoothing capacitance but do not use too much to be safe. Excess smoothing brings other avoidable problems.
Hello RJM and AndrewT, I did not want to move the focus of this thread from the main subject. Very often datasheets are the main source of information and confusion for less skilled DIYers. I have asked that question in good faith. And your responses are great. I am 50 years old music fan who tries to catch up with old hobby.
Thanks
Thanks
rjm said:
I'm sorry Richard, I left this thread a long time ago, did a search now and realized that you already answered this question.
Now I realize that you claim the sound betters with lower voltages, and you may be rigth, but it takes me back to that other question you were asked once:
Have you ever opened up a Phonocube ?
VSPS Update and Phonoclone RC Limited First Photos
Two items today:
1. I have updated the VSPS page. I was concerned about the availability of the RIAA resistors in the 732k and 105k values, so I had a look at what the response would lool like with the next closest standard values, 750k and 110k. The answer is: pretty good. Unless you are going to pay as much attention to the caps (ie measure and match to 1% of the list value) there is really no reason not to use the standard carbon film values.
2. I put together my RC Kit today, and uploaded some photos of the job to my flickr account. Have a look if you like. Everything went smoothly, and it took less than an hour.
/R
Two items today:
1. I have updated the VSPS page. I was concerned about the availability of the RIAA resistors in the 732k and 105k values, so I had a look at what the response would lool like with the next closest standard values, 750k and 110k. The answer is: pretty good. Unless you are going to pay as much attention to the caps (ie measure and match to 1% of the list value) there is really no reason not to use the standard carbon film values.
2. I put together my RC Kit today, and uploaded some photos of the job to my flickr account. Have a look if you like. Everything went smoothly, and it took less than an hour.
/R
Attachments
![1447594068_2afff25839[1].jpg](/community/data/attachments/85/85501-5d66612faa5208ab4c1931dd5b5671b7.jpg)
You mean this? Its also on the VSPS page, btw.
/R
P.S. And before you ask: I know the upper curve looks bad, but it really isnt. (+/- 0.1 dB is bad!? Most of my phonoclones that I have built were worse than this!) You arent going to get anything like the bottom trace anyway unless your capacitors are also matched to under 1%.
/R
P.S. And before you ask: I know the upper curve looks bad, but it really isnt. (+/- 0.1 dB is bad!? Most of my phonoclones that I have built were worse than this!) You arent going to get anything like the bottom trace anyway unless your capacitors are also matched to under 1%.
Attachments
![vsps_response[1].gif](/community/data/attachments/86/86633-77e1cea64e7d190983db9167ad58a17a.jpg)
Hi,
getting the 1uF and three parallel 1uF to match to better than 0.5% is easy. Getting both channels to match is nearly as easy. Getting absolute accuracy is nearly impossible due to inaccuracies in our measuring standards.
If all the caps are matched perfectly but sit 1% low or all sit 1% high, what do these two response curves look like?
Can you eliminate the 0.6db rise @ 20kHz?
How high does that peak go and at what frequency does it level off?
getting the 1uF and three parallel 1uF to match to better than 0.5% is easy. Getting both channels to match is nearly as easy. Getting absolute accuracy is nearly impossible due to inaccuracies in our measuring standards.
If all the caps are matched perfectly but sit 1% low or all sit 1% high, what do these two response curves look like?
Can you eliminate the 0.6db rise @ 20kHz?
How high does that peak go and at what frequency does it level off?
>If all the caps are matched perfectly but sit 1% low or all sit 1% high, what do these two response curves look like?
1% will be trivial deviation. 3% willl be about 0.1 dB, see attached for 732k, 105k, 1.03nF and 3.09nF...
>Can you eliminate the 0.6db rise @ 20kHz?
No, not with the non-inverting topology. You can shift it to higher frequencies somewhat though. As explained on the web page, I have set it deliberately to follow the Allen Wright modified RIAA response.
>How high does that peak go and at what frequency does it level off?
Since the curve is referened to the RIAA response its not a peak, its a levellling off of the RIAA high frequency cut. It continues until the op amp runs out of bandwidth.
1% will be trivial deviation. 3% willl be about 0.1 dB, see attached for 732k, 105k, 1.03nF and 3.09nF...
>Can you eliminate the 0.6db rise @ 20kHz?
No, not with the non-inverting topology. You can shift it to higher frequencies somewhat though. As explained on the web page, I have set it deliberately to follow the Allen Wright modified RIAA response.
>How high does that peak go and at what frequency does it level off?
Since the curve is referened to the RIAA response its not a peak, its a levellling off of the RIAA high frequency cut. It continues until the op amp runs out of bandwidth.
