Many, many years ago I built a JC-2 clone based on the schematics in the "Audio Amateur". Since then I have learned a few things, including how to run HSPICE. For those of you who have not yet learned this tool, I highly recommend it if you ever plan on doing any design.
I am desiging a preamp that is conceptually a follow-on the the JC2, in that it will use discrete op amps. Since I have a large LP collection, it will include a MC pre-preamp stage as well as a line amp stage.
Here is a brief preview of the design.
* Reed relays for all signal switching
* Satellite power supply to minimize magnetic interference
* MC-preamp with En less than that of a 100 ohm resistor
* MC preamp O/L BW flat to >100 KHz, C/L BW flat to >1 MHz
* Teflon caps in signal path , including RIAA network
* 1% RN55 metal film resistors throughout
* MC preamp distortion < noise floor up to onset of clipping
* Line amp capable of driving 100 ohm load
* Line amp distortion below noise floor
* Ability to drive up to +/- 20V peak
I have already simulated the MC preamp, and with the use of low noise matched PNP/NPN devices, there appears to be no problem meeting the noise floor. Additional En reduction may be possible by paralleling input devices (En is reduced by RSS sum).
The next step is the line amp. Since the line amp's source impedances (volume control, for example) are in the 5-10Kohm range, it is right on the cusp between BJTs and JFETs in terms of noise floor at the input stage. I will need to evaluate both types of devices to determine the best choice.
Stay tuned...
JCM
I am desiging a preamp that is conceptually a follow-on the the JC2, in that it will use discrete op amps. Since I have a large LP collection, it will include a MC pre-preamp stage as well as a line amp stage.
Here is a brief preview of the design.
* Reed relays for all signal switching
* Satellite power supply to minimize magnetic interference
* MC-preamp with En less than that of a 100 ohm resistor
* MC preamp O/L BW flat to >100 KHz, C/L BW flat to >1 MHz
* Teflon caps in signal path , including RIAA network
* 1% RN55 metal film resistors throughout
* MC preamp distortion < noise floor up to onset of clipping
* Line amp capable of driving 100 ohm load
* Line amp distortion below noise floor
* Ability to drive up to +/- 20V peak
I have already simulated the MC preamp, and with the use of low noise matched PNP/NPN devices, there appears to be no problem meeting the noise floor. Additional En reduction may be possible by paralleling input devices (En is reduced by RSS sum).
The next step is the line amp. Since the line amp's source impedances (volume control, for example) are in the 5-10Kohm range, it is right on the cusp between BJTs and JFETs in terms of noise floor at the input stage. I will need to evaluate both types of devices to determine the best choice.
Stay tuned...
JCM
Upupa,
The ability to drive 100 ohms should be more than sufficient. A typical preamplifier load consists of the resistive input impedance of the amplifier (typically 20 K ohms) plus the paralled capacitance of the amplifier's input devices and the capacitance of the cable,. Most small signal transistors have Cb < 100 pf. If we assume a common coax such as RG58U, it has a capacitance of 100 pf/meter. The transistors, therefore contribute a capacitance equal to approx 1 meter of additional coax. So, even if we were to drive 10 meters of cable (a lot for a home stereo system), the capacitive reactance Xc would be approx 7K ohms for a 20 KHz signal. Xc = 1/(2*pi*C*f). Add to that the paralleled amp DC resistance, and you get a load of approx 5K ohm. If we repeat this procedure for 20 Hz, we get an load impedance of approx 20K ohms.
There is an insertion loss difference between 20 Hz and 20 KHz caused by the variation in Xc with frequency, but it is very small. You can calculate it by considering the ratio of voltage dividers at 20 Hz and 20 KHz. (100 ohms driving 5K vs. driving 20K). If we use a DC Rl of 20K, then the insertion loss difference between 20 Hz and 20 KHz is 0.13 dB, which is below audability, at least for me.
The ability to drive 100 ohms should be more than sufficient. A typical preamplifier load consists of the resistive input impedance of the amplifier (typically 20 K ohms) plus the paralled capacitance of the amplifier's input devices and the capacitance of the cable,. Most small signal transistors have Cb < 100 pf. If we assume a common coax such as RG58U, it has a capacitance of 100 pf/meter. The transistors, therefore contribute a capacitance equal to approx 1 meter of additional coax. So, even if we were to drive 10 meters of cable (a lot for a home stereo system), the capacitive reactance Xc would be approx 7K ohms for a 20 KHz signal. Xc = 1/(2*pi*C*f). Add to that the paralleled amp DC resistance, and you get a load of approx 5K ohm. If we repeat this procedure for 20 Hz, we get an load impedance of approx 20K ohms.
There is an insertion loss difference between 20 Hz and 20 KHz caused by the variation in Xc with frequency, but it is very small. You can calculate it by considering the ratio of voltage dividers at 20 Hz and 20 KHz. (100 ohms driving 5K vs. driving 20K). If we use a DC Rl of 20K, then the insertion loss difference between 20 Hz and 20 KHz is 0.13 dB, which is below audability, at least for me.
Hi,Upupa Epops said:
I too think that 100ohm driving capability is adequate for a pre-amp.
It will allow direct drive of 600ohm inputs as well as all those higher than that.
It implies a source impedance of under 10ohms, again exemplary.
The output power requirement (4Wpk) will need medium power output transistors in the range 6W to 10W, what do you have in mind?
Upupa,
can you explain why we should need even more current ability out of a pre-amp? (+-20Vpk into 10r = a 20W ClassA power amplifier)
By our experiences ( PMA, me and most of Czech DIY comunity ) is very low output impedance ( less than ten Ohms ) mandatory for really good listening... BUF 634 ( or similar ones ) or more precisely discrete one ( like WJ's diamond buffer ), connected into feedback loop of opamp give best results. Also TPA 6120A is excellent....Or you can make it all discrete, but very low output impedance is mandatory, belive or not...
Hi,
+-20Vpk into 100r is 200mApk.
where's the problem?
Are you confusing load impedance with output impedance?
But Upupa said
So I say again.
A stage designed to drive 100ohms to +-20Vpk should do an excellent job of driving all inputs of >=600ohm.
Give me a reason why the output stage should be designed to drive +-20Vpk into 10ohms.
+-20Vpk into 100r is 200mApk.
where's the problem?
Are you confusing load impedance with output impedance?
this is clearly output impedance and with this Rs one can drive >=100r if the output stage can pass sufficient current.
But Upupa said
which to me means load impedance and in reply followed up with
where Upupa is referring to Rs (output impedance).
So I say again.
A stage designed to drive 100ohms to +-20Vpk should do an excellent job of driving all inputs of >=600ohm.
Give me a reason why the output stage should be designed to drive +-20Vpk into 10ohms.
I tend to agree emotionally but I have also made it easy to test 10 or 110 ohms output impedance of some of my headphone amps. The difference with my 150 ohms Sennheiser HD545 is very small.... but the sound quality of TPA6120 is excellent and also my discrete version is excellent.Upupa Epops said:
Distortion figures
Preamp Output Impedance
One more comment. There is a big difference between an opamp's actual output impedance (Zout) and its ability to drive some maximum signal into a given load. For those interested in a complete explanation, consult Horowitz and Hill, "The Art of Electronics" Second Ed , pp 191.
Zout of an opamp is defined by its OPEN LOOP impedance times the ratio of open loop/closed loop gain. For the design I am considering, its open loop Zout is approx 100 ohms. Assuming we use a conservative amount of feedback (say 40 dB), then the output impedance is 100 ohms/40dB, or 1 ohm. The opamp is designed to have a constant open loop gain over the entire audio range, so Zout is constant over that range.
The above impedance assumptions hold true as long as the opamp can source sufficient current to achieve a full swing. If not, then the swing will be reduced. So what really matters is the maximum output current, not Zout. When I stated earler that the output can drive a 100 ohm load, I was really saying that its output can source +/-20V into 100 ohms = +/- 200 ma peak. This current is wll within capability of the low Vgs MOSFETS used as output source followers.
One more comment. There is a big difference between an opamp's actual output impedance (Zout) and its ability to drive some maximum signal into a given load. For those interested in a complete explanation, consult Horowitz and Hill, "The Art of Electronics" Second Ed , pp 191.
Zout of an opamp is defined by its OPEN LOOP impedance times the ratio of open loop/closed loop gain. For the design I am considering, its open loop Zout is approx 100 ohms. Assuming we use a conservative amount of feedback (say 40 dB), then the output impedance is 100 ohms/40dB, or 1 ohm. The opamp is designed to have a constant open loop gain over the entire audio range, so Zout is constant over that range.
The above impedance assumptions hold true as long as the opamp can source sufficient current to achieve a full swing. If not, then the swing will be reduced. So what really matters is the maximum output current, not Zout. When I stated earler that the output can drive a 100 ohm load, I was really saying that its output can source +/-20V into 100 ohms = +/- 200 ma peak. This current is wll within capability of the low Vgs MOSFETS used as output source followers.
The "what output impedance" question is one that's been going around in my head lately. This morning I did some measurements that might be useful in answering it. I was using a passive preamp for some tests (I don't usually use that one), and was really shocked at how badly some interconnects affected the sound. My thought was that you have to understand the power amp input before you can decide what the preamp output needs to look like. Long story short, I did some impedance measurements of my amp input at various frequencies and under various loads. The conclusion was that the input impedance was mostly accounted for by the passives on the base of the first diff amp transistor, though it did change slightly with both the load on the amp, and the drive level. In my case, for an old copy of a SWTP Tiger amp, that's 2.2K going in, with 20K and 220pF to ground at the base of the transistor. Remember that everything before the base of the transistor is outside the feedback loop, and affects the response of the amp. If you simulate it at 10khz (or foolishly measure it like I did) you get an impedance triangle where R=20K, Z=21K, and X=-6650 (I've rounded a bit). The phase angle was about 18 degrees, but it can be as bad as 25 degrees or more, depending on the cable feeding it. At low frequencies, you can consider it purely resistive. IMO, the preamp output impedance needs to be low enough so there won't be a problem with high capacitance cables and capacitance on the amp input, but high enough to avoid ringing and noise peaking if you connect to something with very low DF coupling caps like polystyrene or polypropylene. I settled on 100 ohms for my active preamp, but don't see any problem going as low as 20 or 30. BTW, if you're going to the trouble of Teflon caps, why not use bulk metal foil resistors? Compared to the caps, they aren't that expensive. I don't believe exotic parts make much audible difference, but it just seems if you're going to go to the trouble and expense, everything should be on the same level 
In the early stages of this thread it seems that there was a misunderstanding between people as to what the thread author meant by 100 ohm capability.
It seems as if some thought he meant 100 ohm output impedance, while others thought he meant ability to drive into 100 ohm input impedance of the next stage.
That misunderstanding seems to have thrown the thread off track, which is a shame.
So, analog_guy, your project sounds very promising and I am sure others as well as myself would be interested to learn of any progress? Is it going well?
It seems as if some thought he meant 100 ohm output impedance, while others thought he meant ability to drive into 100 ohm input impedance of the next stage.
That misunderstanding seems to have thrown the thread off track, which is a shame.
So, analog_guy, your project sounds very promising and I am sure others as well as myself would be interested to learn of any progress? Is it going well?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.