I'm planning to make a discrete audio buffer to interface the LM1036 audio control and a TDA2030 amplifier. The LM1036 cannot drive 16Ohm or 32Ohm headphones by itself, so an audio buffer is required at the output. If no headphones are connected, the buffer will drive a TDA2030 power amp.
I was planning to do this with discrete components, since DIP packaged audio buffers are very rare. I was planning to use a simple common drain amplifier with only one FET per channel, but I don't know what resistor value should I use.
Any ideas?
I was planning to do this with discrete components, since DIP packaged audio buffers are very rare. I was planning to use a simple common drain amplifier with only one FET per channel, but I don't know what resistor value should I use.
Any ideas?
cumesoftware said:
How about 32ohms to match the headphones ?
There are numerous ways to drive headphones, I have even done it cheap with op-amps with a resistor in series with the output. But this was a low tech solution.
You could use a class b driver if you arent bothered about cross over distortion, this would use less current than a class A solution.
What about the LM380 audio driver ?
unclejed613 said:
Yeah.
I have seen for example OPA2134, unity gain stable
being used this way.
Parallell both halves in a dual OP-amp, To drive HeadPhones.
The output current in OPA2134, for example,
is most times more than enough to drive 16-32 Ohms H-Phones.
Something like 30-40 mA can be delivered, from EACH (making Upper limit like 60-80 mA for one pair)
and you should not need use higher Voltage supply, than maybe +-12 VDC.
There are several other OP-Amps with good output current.
When operating at gain = 1, they will drive any modern 16-32 Ohm Headphones.
Because, usually, these headphones need no more than ~ 0.4 - 0.8 Vrms
to make you Half-Deaf .. on both ears
lineup said:
Because, usually, these headphones need no more than ~ 0.4 - 0.8 Vrms
to make you Half-Deaf .. on both ears
I used a couple of op amps to drive the headphones on my disco and it was easily loud enough even with the disco running at 200WRMS. In fact I ran one record deck on one channel and the other on the other channel. This meant I could hear what was playing and what was being cued at the same time.
I've considered using the OPA2134, but then I saw that the output voltage might vary with the load, because of that output resistor. Wouldn't that be a problem?
I also considered using the LM4880, but the THD is too high for an intermediate stage. Only the power stages driving the amps should have a THD of that order of magnitude. So I had to bury this last idea.
Back to the OPA2134. Will it drive 16Ohm? Some headphones have that impedance. I don't want to resort to paralleling op-amps, to save space. Is there other solution? I see that they can supply 35mA, which seems to be more than enough. The problem is that some designs have that nasty resistor at the output to avoid RF interference. Can I take it out in case of unity gain?
http://www.headwize.com/projects/cmoy2_prj.htm
http://www.elecfree.com/electronic/pocket-headphone-amplifier-by-opa134/
I also considered using the LM4880, but the THD is too high for an intermediate stage. Only the power stages driving the amps should have a THD of that order of magnitude. So I had to bury this last idea.
Back to the OPA2134. Will it drive 16Ohm? Some headphones have that impedance. I don't want to resort to paralleling op-amps, to save space. Is there other solution? I see that they can supply 35mA, which seems to be more than enough. The problem is that some designs have that nasty resistor at the output to avoid RF interference. Can I take it out in case of unity gain?
http://www.headwize.com/projects/cmoy2_prj.htm
http://www.elecfree.com/electronic/pocket-headphone-amplifier-by-opa134/
cumesoftware said:
1. Not a problem.
At least in the several projects done this.
2. Don't know.
But any circuit or chip, that had high THD at Gain=1 (buffer only, no voltage gain)
shouldnt be used in Audio.
3. Yes.
Pure math says so.
I said this: 0.4 Vrms in to 16 ohm will make you half deef, on both ears.
400 mVrms = 566 mVpeak
566/16 ohm = 35 mA peak = 25 mA average (RMS)
Output of one ( 1/2 ) OPA2134 is something like 30-40 mA.
Sorry, I was editing:
The problem is that some designs have that nasty resistor at the output to avoid RF interference. Can I take it out in case of unity gain?
Here are the designs I saw. The R5 position varies. I would prefer the later, so R5 doesn't atenuate with the headphones:
http://www.headwize.com/projects/cmoy2_prj.htm
http://www.elecfree.com/electronic/...fier-by-opa134/
The problem is that some designs have that nasty resistor at the output to avoid RF interference. Can I take it out in case of unity gain?
Here are the designs I saw. The R5 position varies. I would prefer the later, so R5 doesn't atenuate with the headphones:
http://www.headwize.com/projects/cmoy2_prj.htm
http://www.elecfree.com/electronic/...fier-by-opa134/
cumesoftware said:
if you want to do it discrete, with JFET and/or lownoise transistors,which is certainly more FUN
than doing Op-Amps 'dead chips',then I suggest:
1. http://www.diyaudio.com/forums/search.php
Use search words like: buffer, discrete, jfet
I have read and posted in maybe 50 topics at this forum.
Where people have built Hi-Fi Buffers.
Both for HeadPhones and Line-amplifier
and using both JFET, lownoise bipolar and MOSFET for their designs.
2. Wait until Nelson Pass publish his series of buffers / preamps.
He has talked a lot of this. He is doing these articles right now.
We are many that await
for The One of Several diyAudio Masters new article works.
Is there any difference on using R5 inside the feedback loop or in series with the load? I prefer the first approach, as the output will be rock solid. However, I've heard that RF might be a problem. Am I right?nigelwright7557 said:
The links I've posted point for these different topologies. I need to know the pros and cons between those two topologies.
It seems the image of the second link got deleted (these incompetent webmasters). I've attached it.
I won't use R3 and R4 will be a short circuit. If I use R5 inside the feedback loop, it should stay rock solid, right? Or should I use R5 outside?
I won't use R3 and R4 will be a short circuit. If I use R5 inside the feedback loop, it should stay rock solid, right? Or should I use R5 outside?
Attachments
And what value of R5 should I use? I'm afraid that if I use a value greater than 10Ohm, the attenuation will be quite noticeable. Unless I compensate with the gain. But then I would have to level the output stage gain.
I see that I can DC decouple the headphones instead of using a resistor. I just saw this in appendix 1 of the Headwize Cmoy page. I guess this might be a better solution. However, I doubt that a polarized electrolytic capacitor can hold this, since the output can go to the negatives as well as to the positives. I guess that solution implies that I should use a non-polarized cap or a polarized cap with the headphones common to the negative rail. Is that right?
I see that I can DC decouple the headphones instead of using a resistor. I just saw this in appendix 1 of the Headwize Cmoy page. I guess this might be a better solution. However, I doubt that a polarized electrolytic capacitor can hold this, since the output can go to the negatives as well as to the positives. I guess that solution implies that I should use a non-polarized cap or a polarized cap with the headphones common to the negative rail. Is that right?
Well. I think there are too many cons on the chip amp approach. I think I should give the discrete a go.
I'm planning to use a JFET with common drain topology. However, I don't know how to calculate the source resistor values, and how to calculate the gate voltage divider resistors as well. I think I should consider the voltage of the divider a voltage such the JFET will drain a near minimum current. Perhaps, it should be set to at least the peak input voltage relative to ground. Right?
I'm planning to use a JFET with common drain topology. However, I don't know how to calculate the source resistor values, and how to calculate the gate voltage divider resistors as well. I think I should consider the voltage of the divider a voltage such the JFET will drain a near minimum current. Perhaps, it should be set to at least the peak input voltage relative to ground. Right?
Found a useful resource:
http://www.muzique.com/lab/buffers.htm
It seems I can use this with bipolar transistors. Since my audio signal is below 0.7V, it won't be heard if I don't use a base divider. I need to use a voltage divider so I can set the base voltage to an average of 1.2V. I will try this tomorrow, and tell you the results.
http://www.muzique.com/lab/buffers.htm
It seems I can use this with bipolar transistors. Since my audio signal is below 0.7V, it won't be heard if I don't use a base divider. I need to use a voltage divider so I can set the base voltage to an average of 1.2V. I will try this tomorrow, and tell you the results.
cumesoftware said:
yes. Was a good webpage you found.
Now if you use this suppose we have a JFET!!!
.... you will get an Input impedance of ~ 100 kohm
( 220k // 220k = 110k )
Then you adjust R2, to a value that is within your JFET current level.
Say you have a JFET which measures max IDSS = 10 mA (with Gate connected to Source).
Then you select R2 (at about 50% of Supply Voltage)
to let JFET run at 5 mA.
For example 12 Volt supply (same as +- 6 V)
We have like 6 Volt across R2.
6V / 1k2 = 5 mA
So we make R2 = 1k2
------------------------------------------------------
This one will have input impedance ~ 500 kohm
actually since R5 is inside the feedback loop, you will not lose any signal (looking at the second citcuit wgere R5 is inside the feedback loop). with a 50 ohm resistor there and a gain of 10, you only get a 5 ohm or less)increase in output impedance. this is because the feedback samples the signal AFTER the resistor.
I must confess that I have my issues with FETs (MOSFETs, JFETS, etc) since they are sensitive to static electricity. The LM1036 has an output impedance of about 20Ohm. So why not use bipolars instead of JFETs? The gain doesn't need to be that great. I think that JFETs are overkill and hard to get. However, I'm open for reasons why should.lineup said:
yes. Was a good webpage you found.
Now if you use this suppose we have a JFET!!!
.... you will get an Input impedance of ~ 100 kohm
( 220k // 220k = 110k )
Then you adjust R2, to a value that is within your JFET current level.
Say you have a JFET which measures max IDSS = 10 mA (with Gate connected to Source).
Then you select R2 (at about 50% of Supply Voltage)
to let JFET run at 5 mA.
For example 12 Volt supply (same as +- 6 V)
We have like 6 Volt across R2.
6V / 1k2 = 5 mA
So we make R2 = 1k2
------------------------------------------------------
This one will have input impedance ~ 500 kohm
There is some confusion here. Putting R5 inside the feedback loop was my inclination, since the amp will regulate the gain according with the load, and so the output would be rock solid. But thenunclejed613 said:
nigelwright7557 suggested that R5 should be in series with the load but outside the feedback loop. That is a problem.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.