Now I've successfully finished a few other smaller audio projects on here I've posted about (unbalanced > balanced stereo converter, active subsonic filter, etc) it's time to start planning for the next big thing I've been asked to build - two or three stereo amps around the 50w mark to feed some high quality 8 ohm speakers.
For this I've gone with a LM3875 Gainclone copy (two mono boards each amp), with a regulated power supply fed by a high-quality toroid. Budget per amp is ~$350 each, which off a rough estimation I will be able to come in under (especially given I already have a heap of high quality caps, metal film resistors and most generic components - RCA/IEC jacks, speaker binding posts, etc).
First design step is the regulated power supply. Trawling through the internet I've come across a few different examples and decided on a circuit using two LT1083's, fed by a MUR860 bridge. It's probably not the cheapest or easiest way to do it, but all the parts are available locally. I'm using a 2x25v toroid, either 300 or 500VAC (thinking I will go with 300 as 500 is probably overkill).
Can anybody see any issues with this circuit?

Now for the PCB, I've got access to a desktop CNC so I am going to attempt to cut it with that. Tracks are super wide (1mm), as they will lose width with a v-bit. The board has been ground filled for ease of cutting.

Any feedback or suggestions so far?
For this I've gone with a LM3875 Gainclone copy (two mono boards each amp), with a regulated power supply fed by a high-quality toroid. Budget per amp is ~$350 each, which off a rough estimation I will be able to come in under (especially given I already have a heap of high quality caps, metal film resistors and most generic components - RCA/IEC jacks, speaker binding posts, etc).
First design step is the regulated power supply. Trawling through the internet I've come across a few different examples and decided on a circuit using two LT1083's, fed by a MUR860 bridge. It's probably not the cheapest or easiest way to do it, but all the parts are available locally. I'm using a 2x25v toroid, either 300 or 500VAC (thinking I will go with 300 as 500 is probably overkill).
Can anybody see any issues with this circuit?

Now for the PCB, I've got access to a desktop CNC so I am going to attempt to cut it with that. Tracks are super wide (1mm), as they will lose width with a v-bit. The board has been ground filled for ease of cutting.

Any feedback or suggestions so far?
Last edited:
Alright, first issue. Turns out although the LT1083 is still listed, I can't grab them. I thought I would swap them out for a LT1584 - similar specs, but different output voltage.
To ensure I haven't done something wrong, Vout is calculated by VREF (1 + R2/R1) + IADJ (R2), yes? That being said, with a R1 value of ~100 ohms, I need ROUGHLY 2k3ish for a 30V output, with an input voltage of 35V (25x1.41)? Does this all seem right?
To ensure I haven't done something wrong, Vout is calculated by VREF (1 + R2/R1) + IADJ (R2), yes? That being said, with a R1 value of ~100 ohms, I need ROUGHLY 2k3ish for a 30V output, with an input voltage of 35V (25x1.41)? Does this all seem right?
Here is the one I use for the LT1083. I get them on Ebay. The boards have been very reliable and adjustable.
Good luck with the project.
LT1083 Adjustable Regulated Power Supply Module DIY Kit Be | eBay
Good luck with the project.
LT1083 Adjustable Regulated Power Supply Module DIY Kit Be | eBay
Here is the one I use for the LT1083. I get them on Ebay. The boards have been very reliable and adjustable.
Good luck with the project.
LT1083 Adjustable Regulated Power Supply Module DIY Kit Be | eBay
Thanks for the link - I might not use that now, but I dI have a use for something like that in the future.
Nobody else has any input?
The eBay posted kit looks to be a pretty fair deal--I notice it uses the protection diodes anyway, though Linear says you don't need them. Don't think you could build it much cheaper than that, with the heat sink and all.
The eBay posted kit looks to be a pretty fair deal--I notice it uses the protection diodes anyway, though Linear says you don't need them. Don't think you could build it much cheaper than that, with the heat sink and all.
Very valid point. After pricing up all components for the PSU plus the amp (not having any cost on the boards), it ends up being much cheaper to buy two of those power supply boards PLUS a chip amp.com that it does buying just components in Australia (despite having all basic caps and resistors already). Thanks import costs!
I ended up buying the above for around $100 delivered, and will just mix and match components until I get what I want.
I just found a big stash of Elna caps, so might try and include them in the build.
I did notice that the eBay kit shows 1,000uF on the schematic but 4,700uF in the picture. In any case, I think it would be prudent to increase it to 10,000uF/35v if it will fit.
I was just reviewing the LT1083 datasheet from Linear Technologies. I had not noticed before, but it CLEARLY states that the MAXIMUM input voltage is 30v, which will limit the output voltage to ~ 29 volts. That's not enough to take full advantage of the LM3875 chip---it needs +/- 35 volts. I wonder if there is another regulator that will do this??
That is the maximum voltage differential from input to output. The input voltage can be higher.[/QUOTE
The data sheet says " the devices are guaranteed
to withstand transient input voltages up to 30V. For input voltages
greater than the maximum operating input voltage some degradation of
specifications will occur."
This seems to be regardless of the input-output difference, doesn't it?
I do not see that in my datasheet.
http://cds.linear.com/docs/en/datasheet/108345fh.pdf
This one only quotes the I/O differential.
Since the reference terminal is lifted above ground by R2, the device does not have the full input voltage across it. You can change R1 and R2 to have the input to reference terminals no higher than 30V and still get higher output voltage since the dropout is under 2VDC. You can also use a zener diode to lift the adjustment pin as long as the Iadj will be enough current to get the zener to turn on. You can bypass the zener with an electrolytic cap to keep the noise down.
http://cds.linear.com/docs/en/datasheet/108345fh.pdf
This one only quotes the I/O differential.
Since the reference terminal is lifted above ground by R2, the device does not have the full input voltage across it. You can change R1 and R2 to have the input to reference terminals no higher than 30V and still get higher output voltage since the dropout is under 2VDC. You can also use a zener diode to lift the adjustment pin as long as the Iadj will be enough current to get the zener to turn on. You can bypass the zener with an electrolytic cap to keep the noise down.
Yeah, I don't see it in THAT datasheet, either; but I DO see it in this one:
http://cds.linear.com/docs/en/datasheet/1083ffe.pdf
Hmmmm.......wonder what the difference is??
http://cds.linear.com/docs/en/datasheet/1083ffe.pdf
Hmmmm.......wonder what the difference is??
I also was wondering exactly WHAT improvement would be gained by having a regulated power supply for a gainclone amplifier? I don't recall seeing this in any of the other designs I've seen...???
On the fixed output LT1083's you ground the adjustment terminal and have a fixed output. The variable version you lift the adjustment pin via a resistor or Zener as part of the voltage reference for the voltage you choose. That is why you can go above 30VDC on the input. The adjustable units need two resistors or a resistor and zener for setting the output voltage.
I also was wondering exactly WHAT improvement would be gained by having a regulated power supply for a gainclone amplifier? I don't recall seeing this in any of the other designs I've seen...???
Consistent, stable voltage that doesn't lag on loud passages. I came across a dozen with regulated supplies when looking at other builds.
Re: the 30v limit - I did see that same comment on the datasheet and wasn't sure what to think about it. A few people have gotten this working with 30v, so I guess I'll just see.
Worse comes to worder I'll just run it at 29v or use the other regulater board I have designed for higher voltages
I also was wondering exactly WHAT improvement would be gained by having a regulated power supply for a gainclone amplifier? I don't recall seeing this in any of the other designs I've seen...???
You won't hit the data sheet performance of the LM3886 without a regulated supply. That's due to the power supply rejection ratio of the LM3886.
There are two ways around this:
1) Use a regulated supply
2) Use a composite amplifier architecture
I chose Option 2 for the Modulus-86, but either approach is valid.
You can see the THD for the LM3886 in an optimized layout with a regulated and unregulated supply in the attached plot. The THD was measured with an Audio Precision SYS-2700.
Tom
Attachments
Why have LT used the same device label (LT1083) for two different regulators?
They have an LT1083 datasheet for the 1083ffe that is the fixed voltage version and a different LT1083 datasheet for the 108345fh that is the adjustable version.
No wonder Dotneck is confused.
They have an LT1083 datasheet for the 1083ffe that is the fixed voltage version and a different LT1083 datasheet for the 108345fh that is the adjustable version.
No wonder Dotneck is confused.
Im interested in using a regulated power supply too. In the doc "AN-1849 An Audio Amplifier Power Supply Design" which is listed together with the lm3886 tech doc, it says:
"For an audio power amplifier, regulated supplies
will need high bandwidth for good audio performance.
The complexity and cost for such a power
supply design may not be acceptable. Most linear regulator
ICs do not have high bandwidth and are slow compared to
audio signals that can result in reduced audio performance."
So Im wondering if LT1083 is good enough? If not, which IC?
I've been thinking of using discrete designs, maybe AMB's sigma 22, The σ22 regulated power supply , but I dont have 10v to drop.
Any other good discrete designs?
Also, is it worth it? Is the difference in THD as in tomchr's plot above audiable?
"For an audio power amplifier, regulated supplies
will need high bandwidth for good audio performance.
The complexity and cost for such a power
supply design may not be acceptable. Most linear regulator
ICs do not have high bandwidth and are slow compared to
audio signals that can result in reduced audio performance."
So Im wondering if LT1083 is good enough? If not, which IC?
I've been thinking of using discrete designs, maybe AMB's sigma 22, The σ22 regulated power supply , but I dont have 10v to drop.
Any other good discrete designs?
Also, is it worth it? Is the difference in THD as in tomchr's plot above audiable?
Im interested in using a regulated power supply too. In the doc "AN-1849 An Audio Amplifier Power Supply Design" which is listed together with the lm3886 tech doc, it says:
"For an audio power amplifier, regulated supplies
will need high bandwidth for good audio performance.
The complexity and cost for such a power
supply design may not be acceptable. Most linear regulator
ICs do not have high bandwidth and are slow compared to
audio signals that can result in reduced audio performance."
So Im wondering if LT1083 is good enough? If not, which IC?
I've been thinking of using discrete designs, maybe AMB's sigma 22, The σ22 regulated power supply , but I dont have 10v to drop.
Any other good discrete designs?
Also, is it worth it? Is the difference in THD as in tomchr's plot above audiable?
I can't think of any requirement for a regulated supply aside from a high enough current capacity, the correct voltage and small enough ripple.
I'm not exactly a genius with this stuff though
I never recommend any newcomer to try building a regulated PSU+Power Amplifier.
The design of the combination is much more complicated than the design of a discrete power amplifier alone.
It is only to be attempted by real Designers or experienced hobbiests/amateurs who are capable of debugging the combination. The stability of the PSU must suit the amplifier's demands.
The design of the combination is much more complicated than the design of a discrete power amplifier alone.
It is only to be attempted by real Designers or experienced hobbiests/amateurs who are capable of debugging the combination. The stability of the PSU must suit the amplifier's demands.
- Status
- Not open for further replies.
- Home
- Amplifiers
- Chip Amps
- Just another Gainclone Build