LM1875/TDA2050 adding output transistors.

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
You’re making a chip amp way harder than it has to be. The original circuit you posted is the only one of the bunch that is practical, then the Rbe value needs to be 1.5 to 2 ohms to get the transistors to turn on at all.then all it can do is boost the current capability - you are still limited by the supply. And the circuit can be finicky, and in ALL cases gives up the built in overload protections that the chip amp is supposed to provide. Short the speaker wires and kapowie, you buy new TIP’s and maybe the chip.

If you need external transistors to provide voltage AND current gain you really don’t want a power op amp in the front end. There are several topologies using two voltage/current gain stages driven by a normal op amp like the 5532 that are a whole lot easier to get working than what you just proposed. Research the QSC circuits for ideas. And they do make versions that don’t use flying rails.
 
Well guys, I wound a ferrite ring about 1" in diameter with a length of 4 stranded magnet wire for an output transformer just to see if it would work. I connected the end of 1 conductor to the start of another giving double the turns for the input, coming from pin 4 of the LM1875. Other end to ground. Resistance of the full length was just over 4 ohms so okay there. The other 2 conductors I connected to the base and emitter of my 2 transistors. I had to crank the input volume up almost all the way to get full sound from the output via the collectors to my speakers (wired parallel for 4 ohms) pretty damn loud and surprisingly punchy too. I thought I would lose bass response using ferrite instead of laminated steel but the setup actually works! It could sound better but I suppose I wound mimimal turns to get a signal out, and the voltage output is 1/2 of the input but it is something. I have a small steel lamination transformer I may wind up with a more appropriate number of turns later but this has been an interesting venture thus far.
I still plan to try a single ended circuit with single power transistor but I would have to connect one speaker connection to positive rail and the other to the collector of NPN power transistor. I may try that scheme tomorrow.
 
UPDATE:
I upped my voltage on my PSU to 30-0-30 and wound a couple output transformers to isolate the feed to the power transistors. While this method does get output to the speaker up to about 100 watts, the volume of the source must be high enough to deliver a strong signal to the base of the transistors, and even then the output is heavily distorted. I tried adding a resistor or capacitor in line to the base of the transistors hoping that might clean things up a bit but nothing satisfactory there. I could bridge a pair of lm1875 chips to increase voltage into the isolation audio transformer in hopes of getting a cleaner signal to the base of the power transistor but I think the real issue here is crossover distortion.
While I had thoughts about using a single supply and a single power transistor in common emitter configuration to boost the voltage and current to the speaker, I foresee some issues there considering that the input signal sweeps both positive and negative voltages from a common ground. I am not familiar with the exact functions of these chips in single supply configuration, though I did get my O-scope in the mail yesterday. Need to find a power supply for that thing and do some investigating I suppose.
 
Hi , with isolation transformers for output transistors you will hardly get proper DC offset at output ,transformer didn't pass DC , so transistors may even amplify different positive and negative signal polarity especially if no feedback used.
Frequency response will be not good too .
To avoid crossover distortion ,you need to partially open transistors ,then they will amplify even small signals ,diodes and resistors can be used for that. Also need to know about small input resistance of bipolar transistors , so your transformers secondary winding should be wound with heavier wire than primary and have less windings .
I have read about output transformers technology ,when multiple windings are wound at once ,and parts of them connected in series ,making primary ,other connected in parallel ,increasing output current and reducing output resistance.
When you will have oscilloscope ,connect a audio signal generator to input ,pc or even phone with some app can be used to generate 1khz test signal ,load amplifier with dummy load resistor, high power with heatsink ,and watch output with oscilloscope ,also look at power supply voltage amplitude ,maybe output at given supply is already reaching maximum possible minus transistor saturation voltage.
 
Thanks ximikas, I have looked at using diodes to offset the crossover junction but I am thinking that the two separate conductors from the isolation transformer may be preventing any bias corrrection? since the opposite end of each conductor is tied to the respective emitters there is 60 volts directly across the two base pins, and it seems the diodes should bleed a measured amount of voltage in reverse bias mode. I have used a battery to measure the amount of voltage bleeding through diodes in reverse bias direction, and have seen a measured amount of difference in them as they have been used and sometimes abused with excessive heat from a soldering iron or other heat dissipation. (from .04 volts up to .8 or .9 volts on occasion, and sometimed full voltage of the battery as the diode is no good.) Seems like I should find a single diode with a high measured reverse bias voltage as I have noted that more diodes actually DECREASES total voltage of bias offset. I am still toying with the circuit and had a full sound using a couple resistors but of course they got quite hot rather quickly. At any rate I am sure there is a viable outcome of success with this odd topology: chip-audio transformer-power transistors. When I get the circuit to operate properly I will share what I have learned from the venture here, and thanks to all for the input and helpful insight!
 
Okay, I have it working fairly well now, used 68R2W resistors from the rails to the leading ends of the 2 secondaries on my output transformer, and a 5.6k 5W resistor between the 2 base pin connections at the other ends of those secondaries. There is a bit of overlap and perhaps the quiescent current is still a bit high but the sound is full and good all the way to zero volume, which at that point there is a very slight hum. I can still play with the circuit a bit but at this point I will say I am quite satisfied with the outcome thus far. Just a couple tweaks away from the build I have been striving for with the parts I have up to my ears. Output at speaker is 102 watts RMS (clipping) using a single pair of output transistors.
 
I used my multimeter measuring AC at the speaker connection with speaker connected. Had pillows over the speakers to dampen the noise but still disturbed the neighbors for a minute. Measured 20v and the impedance of the load disconnected after the test was 3.9 ohms. 20^2/3.9 is 102.564, am I doing this right? I did not use the hold function which would hold the highest value correct?
 
Paint a picture , how your transistors and transformers are connected. Biasing methods are diferent if using common emitter and emiter follower configuration. I measure power the same way, but need to look with oscilloscope,to ensure no clipping occurs yet, and then measure ac voltage at output. Better use lower frequency , not all voltmetters and dmm measures ac voltage accurately ,if frequency is more than 1khz. Clipped power output has no practical reason, you will not like that distorted output and not listed at that level.
 
I had to use multiple strands of very thin magnet wire wound in parallel to pass 3 amperes, lay into the window flat, and provide enough turns to saturate the very small core on the primaries. I also wound another slightly larger one with a single strand and over 200 primary turns, but had to add a 3.3 ohm, 5 watt resistor to make sure the impedance on the primary winding was mor than 4 ohms. I wound the secondaries both at the same time (bifilar) and made sure the beginnings went to the rails, and the endings to the base of the 2 transistors.
There are many variables that can be altered to achieve better performance with this method for utilizing a simple chip amp to drive a set of output transistors. Different winding ratios, higher impedance on the primaries would allow a higher voltage and less strain on the chip, while using fewer turns for the secondaries stepping down the voltages applied to the base of the transistors. Using diodes for biasing the transistors does not seem to work at all probably due to the 2 driving circuits being completely isolated from one another. The resistors at the rails provide further isolation from the DC voltages thereby allowing a much higher value resistor between the 2 base pins to offset the zero point of the 2 transistors.
When I have some time to use my scope to get a better idea of those transitions I will get some photos. I did at least find a 9 volt wallwart so I can finally actually start using my scope. (Now that I am pretty much finished.)
Anyhow, enjoy!
 

Attachments

  • topology1875tt.jpg
    topology1875tt.jpg
    81.8 KB · Views: 268
I also wound another slightly larger one with a single strand and over 200 primary turns, but had to add a 3.3 ohm, 5 watt resistor to make sure the impedance on the primary winding was mor than 4 ohms.

Why would the impedance on the primary side get below 4ohms? You are just driving a pair of emitter followers, right? It is perhaps closer to 4k than to 4R.
 
Ok, i see now you are using common emitter configuration.You can add small resistance to emitters, like 0.1 or 0.33 Ohms, they will stabilyze bias current a little ,also providing some local feedback, increasing linearity. Also what i see ,is high resistance 68ohms with transformer windings, you should add capacitor 1000uf in parallel each 68ohms. This would pass all output from transformer to transistors bases.
 
One sided crossover distortion

After mounting up all my parts and soldering everything up how I like it, I got my scope and started doing some checking and noticed crossover distortion on only one side of the sine wave. The 82 ohm resistors ( I changed from 68 to 82) are .2 ohms difference and the coil windings themselves slightly differ in impedance. The resistor between the two base pins is now at 9.1K instead of 5.6K and the slight buzz at zero volume is gone.
I started trying different low value resistors on the low side with the scope connected until the sine was pure. The overall impedance from one side to the other is different but the sine is clean, and the amp sounds good. The transformer is a 2:1 ratio step down but there are 2 output windings drawing signal from the single input so the power should be about the same, just lower voltage applied to the base of the power transistors. Now that the circuit is complete and I am satisfied, I will add a fan as the heatsinks do get fairly warm. I will post schematics of my exact circuit when I get a chance.
OVER 100 WATTS from a single lm1875 and 2 power transistors from 30-0-30 PSU! Sounds pretty damn good too! :) I wonder what kind of output could be expected if I put 4 windings on the output of the transformer to drive 4 power transistors in bridge configuration?
Also found another schematic using a transformer for the input, slightly different topology but interesting nonetheless.
 

Attachments

  • transformamp.png
    transformamp.png
    39.2 KB · Views: 251
  • osidxvr.jpg
    osidxvr.jpg
    169.2 KB · Views: 273
Last edited:
Oh, and thanks ximikas for the suggestions, I tried the capacitors in line with the 68R (now 82R) like you suggested and that did not work at all. It seemed like a good idea but i suppose it just separated the signal's reference to the emitter from the base pin.
Also appreciate the suggestion to look at the F6 by Nelson, search forums does not readily find the thread but I eventually tracked it down.
F6 Illustrated Build Guide
 
Once you start throwing driver transformers into the mix - it changes the whole ballgame because now you can get both voltage and current gain out of the added transistor stage. This was very common back in the 70’s, but has fallen out of favor because transformers are expensive compared to chips. A transformer designed for this service cost $20 at Digikey. Making your own certainly saves some bucks - and now you’ve sort of got a vintage design. Emitter degeneration and bias tracking diodes can (and should) be added but you’ll figure that out after it sits there and runs away :). The old vintage designs used same sex transistors but the general principle is exactly the same.

In a way, some modern amps have taken a page from some of these old books. A good many professional PA amps use complementary darlingtons, EF3’s or CFPs in common emitter, and drive the whole works directly of a 5532 op amp. It’s cheap, it has a low overall component count, and it works.
 
I am an automotive technician, but am savvy enough to build computers, repair appliances, and started dabbling in electronics so I am new to the field but was certain there was a way to get a simple op-amp to drive power transistors and get ample power on the output. While nobody responding to my post was real keen on the audio isolation transformer idea, I still had to give it a go. Now that I have the biasing resistors set correctly I wonder if the ferrite ring I started with would sound any better having better geometry and very different frequency response characteristics.
I am loving the idea of using transformers to drive power transistors and have come up with an idea to build a system for my car implementing this technique. I have tons of transformers and can even buy small cores, brand new, for $1 each. I would have to make my own bobbins but not a problem. Building stuff is what I do! lol!
I have noted in other schematics that emitter resistors are necessary when using multiple transistors per rail to drive low impedance loads, but are not necessary when using the single NPN and PNP BJT's.
I used resistors for biasing to cancel the crossover distortion as diodes had no effect. I am guessing that since there is 60 volts across the diode being fully blocked there is no voltage at all allowed to flow in either direction.
My next plan is to wind a transformer with 4 outputs and drive 4 BJT's in bridge mode from a single lm1875. Will be interesting to see the outcome of that venture.
 
Last edited:
As drawn you can’t just stick diodes in the base circuit, no. Because of the reverse bias issue. But if you arrange the bias network as a voltage divider from rail to rail you can. Diode, resistor, diode. Then connect the transformer secondaries between the bias tap points and the base. This is the more traditional arrangement from the old 1970’s circuits. You may need to adjust the turns ratio or add a few ohms in series with each diode but that’s easy.

You will never completely get rid of crossover distortion with this circuit. When you get rid of the crossover artifacts with sufficient bias, you will get gm doubling through the region where both transistors are active. So you’re back to having a form of crossover distortion. It is a non issue with emitter followers, but with common emitter it is real. If the entire stage is inside the global feedback loop it becomes negligible. Any time you add gain inside an amp’s feedback loop, the minimum gain for stability goes up so you have to watch that.

If you bridge two of these, build two complete amps and bridge them. It’s much easier to implement the global feedback.
 
Really like the amp I got out of this venture but had to reach for more with a bridge mode using 4 output transistors from a single lm1875 and output transformer. As the other stage on the BJT's is wired with opposing polarity, I noticed that the sine on the input was becoming extremely distorted with artifacts when the second stage was connected. (From output windings to the opposing BJT's.) Someone mentioned crosstalk and it seems that some opposing signal is being imposed on each of the opposing output stages through the audio transformer. I decided to use separate transformers for the 2 opposing stages of BJT pairs to see if the issue would be resolved. Originally thought my transformer was too big so I wound up a smaller ferrite cored transformer and the sine was perfect until power was applied to the output BJT's.
I have attatched a pic of the test board with the smaller audio transformer in my hand that I was hoping would resolve the issue, but did not. I now have 2 identical audio transformers ready to drive the 2 opposing stages to see if this does the trick.
 

Attachments

  • 002.jpg
    002.jpg
    453.1 KB · Views: 170
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.