Hi,
I'm deciding between building a stereo single supply LM1875 or using a pre-made TPA3255 board. I'll be using a linear 34V 3A power supply and I think the LM is more suitable to my liking (was using 3116 board before and much prefer 3886).
I searched this and other forums, and most are referring to datasheet (https://www.ti.com/lit/ds/symlink/lm1875.pdf) with few variation in input (1u vs 2.2u) and output (2200u vs 4700u) coupling caps, and feedback resistor values. Some add additional filtering power supply cap. No other significant changes. I need PCB layout that include only amp and not the power supply, so I think I'll using perf board instead. Here's the diagram from datasheet.
My question is;
1. I planned to use 4700uF for output cap but not sure about input cap so I'm keeping it at 1uF.
2. For layout, based on stability suggestion on 3886, I should keep R5, R6 and C4, C7 close to the chip? Is it applicable for 1875?
3. What wattage rating should I use for R7 in Zobel network?
4. Will perf board give me problems? (assume good soldering)
5. Is there any more suggest changes or adjustment? Thiele network? I'm not using lower impedance speakers, mostly around 4-8 ohms (I heard that 1875 does not like 4 ohm).
Thanks for suggestion,
AP
I'm deciding between building a stereo single supply LM1875 or using a pre-made TPA3255 board. I'll be using a linear 34V 3A power supply and I think the LM is more suitable to my liking (was using 3116 board before and much prefer 3886).
I searched this and other forums, and most are referring to datasheet (https://www.ti.com/lit/ds/symlink/lm1875.pdf) with few variation in input (1u vs 2.2u) and output (2200u vs 4700u) coupling caps, and feedback resistor values. Some add additional filtering power supply cap. No other significant changes. I need PCB layout that include only amp and not the power supply, so I think I'll using perf board instead. Here's the diagram from datasheet.
My question is;
1. I planned to use 4700uF for output cap but not sure about input cap so I'm keeping it at 1uF.
2. For layout, based on stability suggestion on 3886, I should keep R5, R6 and C4, C7 close to the chip? Is it applicable for 1875?
3. What wattage rating should I use for R7 in Zobel network?
4. Will perf board give me problems? (assume good soldering)
5. Is there any more suggest changes or adjustment? Thiele network? I'm not using lower impedance speakers, mostly around 4-8 ohms (I heard that 1875 does not like 4 ohm).
Thanks for suggestion,
AP
1/ The input impedance is 22k so a 1uF will give a -3db point of 7Hz which is fine. It has a higher roll off point than the speaker coupling cap and feedback cap which is all to the good. It defines the LF response correctly in other words.
2/ C4 (the 0.1uF) can usefully fitted right across the pins on the back of the board. In others words as close as possible.
3/ R7 never sees much action in a working amp delivering music and even a 0.25 would run cold but we would never actually fit such a small one. A 2 watt carbon or metal film is perfect.
4/ Should be fine as long as you stick to good practice with grounding.
5/ R5 and R6 look a bit high and I'm surprised they don't stick to the same values as the split rail design tbh. True there are no offset issues to worry over but 200k is very high and noise and even HF performance might suffer (imo). Easy to change and experiment though.
2/ C4 (the 0.1uF) can usefully fitted right across the pins on the back of the board. In others words as close as possible.
3/ R7 never sees much action in a working amp delivering music and even a 0.25 would run cold but we would never actually fit such a small one. A 2 watt carbon or metal film is perfect.
4/ Should be fine as long as you stick to good practice with grounding.
5/ R5 and R6 look a bit high and I'm surprised they don't stick to the same values as the split rail design tbh. True there are no offset issues to worry over but 200k is very high and noise and even HF performance might suffer (imo). Easy to change and experiment though.
Thanks all. I understand that gain is R5/R6, which is 200/10 = 20. If u think it is high, what value should I go for R5? I think it can't go below 10 based on spec sheet.
As Mooly said, the absolute values look high, its more typical to see 20k and 1k. I'd keep the ratio at 20:1 at least to begin with. Unless you're feeling bold in which case decrease R5 to 2k with 20k for R6.
Just for fun I simulated the single rail design using an LM1875T model I created a while back. The 1 ohm and 0.22uF Zobel looks... I dunno, a bit severe (but of the course the data sheet is what it is and should be a correct worked example).
Well my model oscillated and the reason was the 1 ohm. Increase it a little (anything past 2.5 ohm in the sim) and all is good. Typical values are usually 10 ohm and 0.1uF. I wonder if I should tweak my model so it is stable with 1 ohm and 0.22u loading.
Anyhow:
Well my model oscillated and the reason was the 1 ohm. Increase it a little (anything past 2.5 ohm in the sim) and all is good. Typical values are usually 10 ohm and 0.1uF. I wonder if I should tweak my model so it is stable with 1 ohm and 0.22u loading.
Anyhow:
Thanks. I also tried but lack LTSpice skill LOL
Am I use correct file? I tried using both PSpice and TINA-TI spice to manually create .asy but both give same error.
Am I use correct file? I tried using both PSpice and TINA-TI spice to manually create .asy but both give same error.
Where did you get your LM1875 model from ?
This is the model I used:
And made it into a model that can be called up and used anytime.
This is the model I used:
And made it into a model that can be called up and used anytime.
The output decoupling caps cripple the sound (you can look at D. Self - Audio Power Amplifier Design Handbook, and many others of course).
In the past I very successfully experimented with LM1875 (8 Ohms speakers) with 40V PS and virtual ground (2x4700uF). No hum, very low noise and the THD went little bit lower than LM1875 with bipolar PS.
In the past I very successfully experimented with LM1875 (8 Ohms speakers) with 40V PS and virtual ground (2x4700uF). No hum, very low noise and the THD went little bit lower than LM1875 with bipolar PS.
Output decoupling caps increase the THD by 10-50x. Just search and read about this subject. There is lot of information.
The single supply version is safer for a DIY project because it protects the speaker from DC. However, it can give a nasty thump when power is applied. Feed the output through a switch or jumper to a resistor or the speaker. Apply power with the resistor engaged for about 3 seconds, then select the speaker. Understand that the capacitor mostly charges through the resistor so that the speaker will not thump.
Thanks a lot for sharing. I downloaded from TI website but it's not compatible with LTSpice.
I think we all know (and were taught) that output cap degrade sound quality. I've built a chip-based with output cap once and only using OCL designs after that. I always used some speaker protection circuit and now it's much easier since there's a lot of available premade board. I might end up adding one in this circuit to prevent power on thump.
Recently, I stumble upon threads on "retro" style amp in solid state forums (e.g. Apex AX6 here and other sites) and found people show passion about the amps, component selection, etc and it does look fun. So, I intrigued to build one but with a chip instead go all solid state for stability and my limited capability in redesigning the circuit.
As for distortion, with larger perfect cap, it shouldn't effect distortion at audio frequency. Given real world use of high quality capacitor, it guess it should be good enough for audio application since there's non linearity in both frequency response and impedance of capacitor and connected speakers. Even with increased distortion, it should be still at acceptable in casual listening.
Recently, I stumble upon threads on "retro" style amp in solid state forums (e.g. Apex AX6 here and other sites) and found people show passion about the amps, component selection, etc and it does look fun. So, I intrigued to build one but with a chip instead go all solid state for stability and my limited capability in redesigning the circuit.
As for distortion, with larger perfect cap, it shouldn't effect distortion at audio frequency. Given real world use of high quality capacitor, it guess it should be good enough for audio application since there's non linearity in both frequency response and impedance of capacitor and connected speakers. Even with increased distortion, it should be still at acceptable in casual listening.
Fair enough 🙂 I haven't looked at those.
On output caps... all I will say is that some of the most highly regarded amps use cap coupling. My view is that its not a problem when it comes to sonics.
Thanks Mooly for sharing the model. It works now. So far, this is the design based on recommendations I got here (Zobel adjustment, supply filtering cap and gain resistors).
I'm considering few more things..
1. R4 value, some people suggest it may a little too high and might cause oscillation (I read some where to use 47k?). Datasheet also suggests an additional HF coupling capacitor between 50-500pf to prevent high frequency oscillation but I don't see this in practice...
2. Some suggest to put a resistor close to 470-500 ohms across speaker terminal to prevent turn on pop while C6 is charged? I'll try that later if the problem comes up.
3. R6/R5 ratio, while it is the same as datasheet, what benefit of having lower absolute value? Is it to increase feedback current for more stability?
AP
I'm considering few more things..
1. R4 value, some people suggest it may a little too high and might cause oscillation (I read some where to use 47k?). Datasheet also suggests an additional HF coupling capacitor between 50-500pf to prevent high frequency oscillation but I don't see this in practice...
2. Some suggest to put a resistor close to 470-500 ohms across speaker terminal to prevent turn on pop while C6 is charged? I'll try that later if the problem comes up.
3. R6/R5 ratio, while it is the same as datasheet, what benefit of having lower absolute value? Is it to increase feedback current for more stability?
AP
R4 is just good practice to ensure the input coupling cap is always discharged if the input is floating. It plays no other part. Similarly adding a resistor as you mention across the speaker output is also good practice as it would charge the speaker cap even with no load connected and prevent a loud thump if you then connected speakers. It only odes any good though if you turn the amp on with no load attached.
Input filtering can be a very good thing as it reduces stray pickup of HF radiated interference and also limits the rise time of any out of band signals that may cause distortion and slew rate limiting.
R5/R6 is usually set to give a decent noise figure, more important with preamp design really but non the less...
A lower value helps with HF response and non linearities due to the chip input capacitance and other input characteristics.
Input filtering can be a very good thing as it reduces stray pickup of HF radiated interference and also limits the rise time of any out of band signals that may cause distortion and slew rate limiting.
R5/R6 is usually set to give a decent noise figure, more important with preamp design really but non the less...
A lower value helps with HF response and non linearities due to the chip input capacitance and other input characteristics.