I am building an electret mic with two elements through a differential opamp. I have TL072s and NE5534s to choose from.
The amps were first tested as followers, with a sine wave tone generated from a speaker near the element. The 072s work good, faithfully representing the input. The 34s however, produce distortion, and noise. There is some noise with no input and with the same input as the 072s, the waveform is distorted. It gets worse at freqs closer to 10kHz, with several spikes in the area with almost the same amplitude on the frequency response graph.
The 072 data sheet suggests using a resistor/cap shunted across the output, which was added. The 34s have the compensatory 22pF cap across the right terminals. The offset null worked but the positive input had to be grounded to make the adjustment.
The resistor/cap pair was added to the 34, as well as a high value load resistor after the blocking cap of the electret, with no discernible improvement.
My questions are:
1. Is it possible the Chinese bootleg NE5534s are defective.
2. The 22uF blocking electrolytics are oversized for the application and the 34s are amplifying any distortion introduced there, due to their sensitivity.
3. The NE5534 inherently needs some kind of a filtering circuit, or certain circuit characteristics, heretofore unknown to this newb.
Any comment would be appreciated.
The amps were first tested as followers, with a sine wave tone generated from a speaker near the element. The 072s work good, faithfully representing the input. The 34s however, produce distortion, and noise. There is some noise with no input and with the same input as the 072s, the waveform is distorted. It gets worse at freqs closer to 10kHz, with several spikes in the area with almost the same amplitude on the frequency response graph.
The 072 data sheet suggests using a resistor/cap shunted across the output, which was added. The 34s have the compensatory 22pF cap across the right terminals. The offset null worked but the positive input had to be grounded to make the adjustment.
The resistor/cap pair was added to the 34, as well as a high value load resistor after the blocking cap of the electret, with no discernible improvement.
My questions are:
1. Is it possible the Chinese bootleg NE5534s are defective.
2. The 22uF blocking electrolytics are oversized for the application and the 34s are amplifying any distortion introduced there, due to their sensitivity.
3. The NE5534 inherently needs some kind of a filtering circuit, or certain circuit characteristics, heretofore unknown to this newb.
Any comment would be appreciated.
do the electrets have built in jfet amplifier/buffer or are you trying to go direct off the capsule
hard to believe anyone could make money counterfiting 50 cent parts
the 5534 is a single, require external compensation parts for gain <~3 - the dual part # is 5532 to be TL072 socket comaptable, still has noticible bias current requirement, current noise that make it undesireable for direct connection to ultra high Z like electret mics
hard to believe anyone could make money counterfiting 50 cent parts
the 5534 is a single, require external compensation parts for gain <~3 - the dual part # is 5532 to be TL072 socket comaptable, still has noticible bias current requirement, current noise that make it undesireable for direct connection to ultra high Z like electret mics
Thank you for responding.
The electrets do, in fact, have jfets. They provide a decent signal straight out.
Could the compensation cap being ceramic cause the distortion?
I don't think I have a handle on bias current. How can I implement some?
Thanks.
The electrets do, in fact, have jfets. They provide a decent signal straight out.
Could the compensation cap being ceramic cause the distortion?
I don't think I have a handle on bias current. How can I implement some?
Thanks.
I'll try to explain bias current by comparing a jfet and bipolar transistor amplifier.
The jefet is a normally on device, so for it to operate as an amplifier it has to be biased off some. This is done by reverse biasing the gate source junction. There is still some leakage current but this is usually femto or pico amps.
A bipolar transistor is normally off, so for it to operate as an amplifier it must be biased on. This is done by forward biasing the base emitter junction. This bias current can be nano or microamperes in a small signal amplifier to many milliamperes in a power amplifier. This bias current has to come from somewhere for the bipolar transistor to operate.
The OP-27 opamp has a bias current compensation circuit so the input transistors do not draw bias current from the signal source. The NE5534 opamp does not have this feature so the input transistors draw their bias current from the signal source.
Ray
The jefet is a normally on device, so for it to operate as an amplifier it has to be biased off some. This is done by reverse biasing the gate source junction. There is still some leakage current but this is usually femto or pico amps.
A bipolar transistor is normally off, so for it to operate as an amplifier it must be biased on. This is done by forward biasing the base emitter junction. This bias current can be nano or microamperes in a small signal amplifier to many milliamperes in a power amplifier. This bias current has to come from somewhere for the bipolar transistor to operate.
The OP-27 opamp has a bias current compensation circuit so the input transistors do not draw bias current from the signal source. The NE5534 opamp does not have this feature so the input transistors draw their bias current from the signal source.
Ray
Thank you for that concise explanation!
OK. Somehow the design requires the compensation capacitor. I guess I will try some larger values since there is nothing else to try. I cannot change the input magnitude.
I also looked at the OnSemi data sheet for NE5534. Those circuits are test circuits. They are adding load and capacitance to the follower. Since National did not indicate that, I assumed the components were required.
Just got done looking on ebay for some chips. I saw inexpensive ones supposedly made by TI. The seller made the mistake of posting a high-resolution picture with zoom-in capability. The State of Texas logo appears very sloppy. Another seller's picture shows a chip with no marking whatsoever. Finally (going higher in price), there are the OnSemi brand for 5 US each plus shipping, in the US.
I will try those and see what happens.
OK. Somehow the design requires the compensation capacitor. I guess I will try some larger values since there is nothing else to try. I cannot change the input magnitude.
I also looked at the OnSemi data sheet for NE5534. Those circuits are test circuits. They are adding load and capacitance to the follower. Since National did not indicate that, I assumed the components were required.
Just got done looking on ebay for some chips. I saw inexpensive ones supposedly made by TI. The seller made the mistake of posting a high-resolution picture with zoom-in capability. The State of Texas logo appears very sloppy. Another seller's picture shows a chip with no marking whatsoever. Finally (going higher in price), there are the OnSemi brand for 5 US each plus shipping, in the US.
I will try those and see what happens.
Thank you for that concise explanation!
OK. Somehow the design requires the compensation capacitor. I guess I will try some larger values since there is nothing else to try. I cannot change the input magnitude.
I also looked at the OnSemi data sheet for NE5534. Those circuits are test circuits. They are adding load and capacitance to the follower. Since National did not indicate that, I assumed the components were required.
Just got done looking on ebay for some chips. I saw inexpensive ones supposedly made by TI. The seller made the mistake of posting a high-resolution picture with zoom-in capability. The State of Texas logo appears very sloppy. Another seller's picture shows a chip with no marking whatsoever. Finally (going higher in price), there are the OnSemi brand for 5 US each plus shipping, in the US.
I will try those and see what happens.
OK. Somehow the design requires the compensation capacitor. I guess I will try some larger values since there is nothing else to try. I cannot change the input magnitude.
I also looked at the OnSemi data sheet for NE5534. Those circuits are test circuits. They are adding load and capacitance to the follower. Since National did not indicate that, I assumed the components were required.
Just got done looking on ebay for some chips. I saw inexpensive ones supposedly made by TI. The seller made the mistake of posting a high-resolution picture with zoom-in capability. The State of Texas logo appears very sloppy. Another seller's picture shows a chip with no marking whatsoever. Finally (going higher in price), there are the OnSemi brand for 5 US each plus shipping, in the US.
I will try those and see what happens.
There is just no point in buying such parts from eBay. 95% of the time, if you buy from a Hong Kong/China seller, you will get counterfeit parts. Other sellers, its about 70% risk. By that time, you may as well have ordered from Digikey.
Update
When I went to compensation cap swap, it was found there was a bad connection on the breadboard.
But there was horrific distortion that came and went! I am using Visual Analyzer with a low-end sound board, so I instinctively blamed the hardware.
Further investigation by moving around the test lead ends near the following device, was found to be the culprit:
When I went to compensation cap swap, it was found there was a bad connection on the breadboard.
But there was horrific distortion that came and went! I am using Visual Analyzer with a low-end sound board, so I instinctively blamed the hardware.
Further investigation by moving around the test lead ends near the following device, was found to be the culprit:
I believe these are knock-offs or out-of-spec rejects from the majors. For speech recognition, micro levels of noise and distortion differences are insignificant. But if I were building a 1,000 USD amp, I would tend to go with the A suffixes from Newark or Digikey. I see that audio can brought to a spiritual level, almost insane, when I look at some of the tube amp descriptions, for instance.
I learned much from this experience. Since the app is mobile, the mic must be shielded. I also see why PCs have those metal shields for the front bays when they are blank. I have not gotten around to walkie talkies and HAMs yet. I wonder what they will do? I recall having heard clear speech from a local HAM operator through stereo speakers when I was younger. If my mic has, say, a gain of 20, I suspect there will be an issue here.
Thank you everyone for your thoughts.
I learned much from this experience. Since the app is mobile, the mic must be shielded. I also see why PCs have those metal shields for the front bays when they are blank. I have not gotten around to walkie talkies and HAMs yet. I wonder what they will do? I recall having heard clear speech from a local HAM operator through stereo speakers when I was younger. If my mic has, say, a gain of 20, I suspect there will be an issue here.
Thank you everyone for your thoughts.
+1 to vac. I buy from Taobao, hundreds of ICs of various kinds - so far haven't found one fake, but do I keep my eyes peeled and don't try buying from the very cheapest sellers who are obviously dishonest.
<edit> jcx's point is a good one, the price of an NE5532 here is 1rmb, that's about 15cents. Where's the margin for fakes?
Last edited:
So you're saying its in Polida's economic interest to trash eBay sellers? If so, that's too much of a coincidence....
So you're saying its in Polida's economic interest to trash eBay sellers? If so, that's too much of a coincidence....
There was nothing wrong with the units. The problem was bad connections and radiation from two screw-in flourescents right on the bench, as stated above.
Was the 5534 properly compensated? They have a tendency to oscillate if not used properly as they are not unity gain stable. You can't use it as a follower without compensation. 5532s are unity stable and have the same pinout as TL072.
G²
G²
That is how the bad connection was found. I thought perhaps the 22uF ceramic was the culprit, so when I changed it to a 27, everything was fine. But I knew something was fishy and went back to the 22 and it was the perfect all of a sudden. I realized then it was a connection.
- Status
- Not open for further replies.