so I have been using a LM1875 for a while now and decided to switch to a BTL setup to boost the ouput from the same voltage.
here is how I have it wired, however I am having an issue that the gain on the inverting amplifier (green) is very frequency dependant, so at 100Hz for instance its output would be 2V p-p, when the non-inverting one is at 10v p-p on a test sine wave.
but at 1k Hz they are both 10v p-p, and at 10k, the inverting amp is clipping a square wave at 22v p-p when the non-inverting one is still happily humming along at 10v p-p output.
the other thing i noticed is the output is not out of phase on the inverting one, it is only like 60 degrees shifted rather than 180 deg. hopefully this makes sense.
this is how In have it wired right now. (forgive the crude print, its 2 am and I have to work tomorrow
here is how I have it wired, however I am having an issue that the gain on the inverting amplifier (green) is very frequency dependant, so at 100Hz for instance its output would be 2V p-p, when the non-inverting one is at 10v p-p on a test sine wave.
but at 1k Hz they are both 10v p-p, and at 10k, the inverting amp is clipping a square wave at 22v p-p when the non-inverting one is still happily humming along at 10v p-p output.
the other thing i noticed is the output is not out of phase on the inverting one, it is only like 60 degrees shifted rather than 180 deg. hopefully this makes sense.
this is how In have it wired right now. (forgive the crude print, its 2 am and I have to work tomorrow
The green 1k needs to be in series with the green 1uf. The unmarked cap at the bottom isn't needed. The green 22k feedback resistor should be 23k to make the gain of the inverting stage equal the gain of the non inverting.
Downsides... the inverting stage has a 1k input impedance. This means it loads the source heavily. The green 1uf input cap needs raising to 100uf to equalise the input cut off frequency between the two stages. Upper left blue 100k needs a cap across it to remove all audio, say 47uf.
Its not a great set up tbh unless you just experimenting.
Downsides... the inverting stage has a 1k input impedance. This means it loads the source heavily. The green 1uf input cap needs raising to 100uf to equalise the input cut off frequency between the two stages. Upper left blue 100k needs a cap across it to remove all audio, say 47uf.
Its not a great set up tbh unless you just experimenting.
You can simplify it a bit - remove the green 1uF. Remove the green cap just above GND and route the dangling end of the 1k to the bottom end of the blue cap above the higher GND. Disconnect said cap from GND.
As Mooly pointed out, the feedback resistors need to be nudged a little to get equal gain from the two halves. The bottom one needs to be 23k when the top's 22k.
As Mooly pointed out, the feedback resistors need to be nudged a little to get equal gain from the two halves. The bottom one needs to be 23k when the top's 22k.
Look up the datasheets for the TDA2040. IIRC, they show a bridge configuration which will work for the 1875. I don't recommend your config without op amp buffers/inverters.
With 8 ohm load, you are safe to operate the 1875 up to 30v supply which includes some headroom for reactive loads. If you decide to try a 4 ohm load, 18v is about as far as you should go.
With 8 ohm load, you are safe to operate the 1875 up to 30v supply which includes some headroom for reactive loads. If you decide to try a 4 ohm load, 18v is about as far as you should go.
Variations of this circuit have been around for a long time. Its performance is dependent on an array of precision matched resistors and oddball values. Any error in the resistor network will degrade performance. I haven't seen any commercial applications of this circuit (common in cheap car radio "power boosters" of yesteryear) that worked very well.
An easy way out would be to use a differential line driver like DRV134. DRV134PA Texas Instruments | DRV134PA-ND | DigiKey The whole resistor network is integrated into the chip and laser trimmed for highest precision and lowest distortion. It provides a fixed gain of 6 dB in differential mode. No external feedback network is required. It does a lot of heavy lifting for its $6.00 price tag.
This could drive non-inverting chip amps. Since it's designed to drive 600 ohm loads it could even drive low impedance input inverting amps if you like. It frees up a lot of wiggle room for your design.
There is a circuit floating around out there that uses this chip in a composite differential chip amplifier, but this greatly complicates your design.
No matter how you go, this chip will make your design easier and better.
An easy way out would be to use a differential line driver like DRV134. DRV134PA Texas Instruments | DRV134PA-ND | DigiKey The whole resistor network is integrated into the chip and laser trimmed for highest precision and lowest distortion. It provides a fixed gain of 6 dB in differential mode. No external feedback network is required. It does a lot of heavy lifting for its $6.00 price tag.
This could drive non-inverting chip amps. Since it's designed to drive 600 ohm loads it could even drive low impedance input inverting amps if you like. It frees up a lot of wiggle room for your design.
There is a circuit floating around out there that uses this chip in a composite differential chip amplifier, but this greatly complicates your design.
No matter how you go, this chip will make your design easier and better.
The inputs don't draw any current (well only microscopic amounts). What draws the current is the configuration the device is used in.
In inverting mode the input signal is applied through a series resistor to the inverting input. That in itself draws no current. What does cause the current flow is the fact the inverting input has another resistor tying it to the output of the chip. This means the inverting input always sees zero volts at the input pin, and as a consequence, the value of the series input resistor now becomes the input impedance which is seen by the driving source component.
ok, thank you mooly. when / if I use this in my project i will bump the values of that divider so the proportion is the same but the values are higher.
I am using this in an industrial environment. so I only have +24V DC power as my supply. with a single 24V rail I was putting out only around 7W to the speaker but for what they were used for it worked well and was simple. the new ones I have to make now have to be significantly louder but I can't completely gut the design for these.
I am also considering having a switch mode inverter to make me a -24V rail and using the single 1875 as well but the it dumps 19W of heat like that and I am running into limitation of a convection mode heatsink to handle that. I may end up doing the BTL and maybe the inverter as well (tuned down to -5 to -12v probably) depending on how the thermal loading ends up with the two 1875s.
the next version I have decided to switch up to a class D amp (aiming for 80W - 100W and the speaker is rated for 125W RMS) but that is a future project. (with a single 30V rail and 90% efficiency it should be around there)
I am using this in an industrial environment. so I only have +24V DC power as my supply. with a single 24V rail I was putting out only around 7W to the speaker but for what they were used for it worked well and was simple. the new ones I have to make now have to be significantly louder but I can't completely gut the design for these.
I am also considering having a switch mode inverter to make me a -24V rail and using the single 1875 as well but the it dumps 19W of heat like that and I am running into limitation of a convection mode heatsink to handle that. I may end up doing the BTL and maybe the inverter as well (tuned down to -5 to -12v probably) depending on how the thermal loading ends up with the two 1875s.
the next version I have decided to switch up to a class D amp (aiming for 80W - 100W and the speaker is rated for 125W RMS) but that is a future project. (with a single 30V rail and 90% efficiency it should be around there)
https://www.sparkfun.com/datasheets/Components/General/STA540.pdf
why not leave the lm1875 and use something more suited ?
why not leave the lm1875 and use something more suited ?
If the OP uses the 24v supply and bridged into a 8 ohm load, the load current will be around 1.6 amps. I measured the 1875 to current limit at 3 amps rms (continuous sine wave). So he has plenty of reactive load headroom. He would be safe at 2 amps with typical speaker loads.
With a single 1875 and 55.2v supply, I measured 33.62 watts before clipping. (2.05 amps into 8 ohms). You can't exceed this using a bridge config and using 8 ohm load because of the output current limitations but can pull it off with less supply volts. So unless you are constrained by the power supply it makes more sense just to use a single IC and higher supply voltage.
Sorry but it will be higher (although still below 3A peak).
In a BTL amp, the speaker sees peaks of +Vcc minus semiconductor drop.
Worst but possible case would be 24V - 4V , so 20V/8r = 2.5A Pk
In fact, you need a little more, because speaker loads are reactive, so add at least 20% more.
So yes, an LM1875 is still usable but just on the edge.
Off a single 24 volt supply, a bridged 1875 ought to be able to deliver 30 watts RMS into an 8 ohm load with little margin to spare. Like mentioned it will be "just on the edge."
It seems like an interesting project. I'm interested in how well it works in the real world.
Edit: This configuration could be run off a large laptop power supply and still pack a punch. The amplifier circuit could be made very small, but a lot of heat sink is still required.
It seems like an interesting project. I'm interested in how well it works in the real world.
Edit: This configuration could be run off a large laptop power supply and still pack a punch. The amplifier circuit could be made very small, but a lot of heat sink is still required.
Last edited:
here is what the earlier versions looked like (v1 is a 4"x4"x2" box) (v2 is a 2" x 3"x2" box which has the same circuitry. but a smaller speaker so not as loud.) v3 is the new one which is 4"x4"x3" box to allow for a "longer" horn style speaker (its a horn tweeter that cant go below 1kHz but for alarm signals it works)
the V1 unit i have about 160 units in the field over the last year and a half.
V1 and V2
V1 speaker I got on buyout special from parts-express for about a buck per speaker but they work really well. I bought almost 120 of them and i have been using them from "stock" ever since, I maybe have 30 left now) but for the specs it is a steal. earlier units had a visaton R10s
parts-express speaker
V1 earlier circuit
here is the PCB i made for the V1 units (there have been several revisions but this is a good shot of one of them.
V3 in progress (that yellow board is the most recent V1 circuit, but I don't have any good pictures of it right now.) the speaker (so far) for the V3 is a Pyle PH25. the big convection heatsink on the top outside of the box.
the V1 unit i have about 160 units in the field over the last year and a half.
V1 and V2
V1 speaker I got on buyout special from parts-express for about a buck per speaker but they work really well. I bought almost 120 of them and i have been using them from "stock" ever since, I maybe have 30 left now) but for the specs it is a steal. earlier units had a visaton R10s
parts-express speaker
V1 earlier circuit
here is the PCB i made for the V1 units (there have been several revisions but this is a good shot of one of them.
V3 in progress (that yellow board is the most recent V1 circuit, but I don't have any good pictures of it right now.) the speaker (so far) for the V3 is a Pyle PH25. the big convection heatsink on the top outside of the box.
I have real measurements, not estimates. With 24v supply, single 1875, non inductive 4 ohm load, The output made 17.875 V pp (6.32v rms or 1.58 A rms) at the point just before clipping. Vs measured at pins 3 and 5 while under test. So in bridge config, with 8 ohm load, double the Vout, you get the same current. (I'm using rms Amps, not peak).
17.875Vpp is exactly equivalent to 8.9375Vpk.
The current demanded by the resistive load is 2.234375 Apk, if the load is exactly 4r0 while the test is carried out.
You don't need to convert anything to rms.
The output power is 2.234375*8.9375/2 = ~9.985W
From that, we get a bridged pair will deliver ~19.97W into 8r0
That's a lot of hardware for ~20W into a resistive load.
We should next consider what happens when a reactive speaker is the load.
I have given my opinion but you don't want to believe.
The current demanded by the resistive load is 2.234375 Apk, if the load is exactly 4r0 while the test is carried out.
You don't need to convert anything to rms.
The output power is 2.234375*8.9375/2 = ~9.985W
From that, we get a bridged pair will deliver ~19.97W into 8r0
That's a lot of hardware for ~20W into a resistive load.
We should next consider what happens when a reactive speaker is the load.
I have given my opinion but you don't want to believe.
Don't want to believe what?
I used rms because that is what I had recorded on my tables.
My point was at 24v there is plenty of headroom for most 8 ohm loudspeaker loads.
I also stated that since you can get the same output from a single chip if you increase the supply, going bridge did not make sense unless you are supply voltage limited.
An externally hosted image should be here but it was not working when we last tested it.
you posted my quote in post11
Told you you didn't want to believe.
All these values are for a purely resistive load. The manufacturers deliberately use resistive loads as well.
You must take account of reactive speaker loads.
You must also take account of effective impedance with fast changing signals.
You are omitting these last two evaluations and as a result coming to the wrong conclusions.
BTW,
those tables seem familiar.
Did I make comment on how to interpret the results back at the time they were posted?
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.