I am an unabashed tinkerer and have been playing with phono preamp designs for the last few months. In the effort to provide the ever important "More knobs and switches" aspect, I've been working on a kitchen sink preamp. The idea is to provide a great deal of control over the circuit through switches and whatnot.
Here's what I have so far:
- Single-stage RIAA decoding using RJM's excellent values
- Selectable gain, loading and capacitance
- A (hopefully) decent power supply
So the questions, beyond anything obvious that I missed...
1. How do the values look for the input cap selector?
2. Did I get the value for R1e and R8e correct? Should it be higher?
3. I supposed I could ditch the output capacitors? Thoughts?
I'm somewhat of a noob at audio DIY, so feel free to slag away!
Here's what I have so far:
- Single-stage RIAA decoding using RJM's excellent values
- Selectable gain, loading and capacitance
- A (hopefully) decent power supply
So the questions, beyond anything obvious that I missed...
1. How do the values look for the input cap selector?
2. Did I get the value for R1e and R8e correct? Should it be higher?
3. I supposed I could ditch the output capacitors? Thoughts?
I'm somewhat of a noob at audio DIY, so feel free to slag away!
Attachments
Well, it has been a day and the experts are leaving this alone. I'll go first & let them pile on. OPA2134 is a lot faster than a 4558, which needs no .1 uf ceramic bypass caps. I would suggest you use them on the power supplies, because wound electrolytic caps have anough inductance to be too slow to keep fast op amps from oscillating. If your 10 uf bypass caps are $7 aerovox ceramics, I withdraw the suggestion, but 220 uf is obviously wound. Also, fast op amps need about 20 pf across the feedback resistor to prevent oscillation. I would put it across the one without series capacitance. OPA is not extremely fast, so maybe 10 pf is enough.
Low voltage switches (under 1 v) are not very reliable due to dust and corrosion. Unless your switch contacts are gold, I would expect no sound or a beep everytime you switch the input. Even the 25 mv tone generator signals have trouble getting across the wirewrap on my Hammond organ preamps. Several wirewraps were soldered by dealer service.
Not sure why people buy this exotic device. It has an insanely low THD spec, but it is at unity gain which is kind of useless. You're not using it at unity gain. Cheap old 33078 is better on noise (4.5vs 8) and has no FET's. Other devices don't have FET sensitivity warnings.
Low voltage switches (under 1 v) are not very reliable due to dust and corrosion. Unless your switch contacts are gold, I would expect no sound or a beep everytime you switch the input. Even the 25 mv tone generator signals have trouble getting across the wirewrap on my Hammond organ preamps. Several wirewraps were soldered by dealer service.
Not sure why people buy this exotic device. It has an insanely low THD spec, but it is at unity gain which is kind of useless. You're not using it at unity gain. Cheap old 33078 is better on noise (4.5vs 8) and has no FET's. Other devices don't have FET sensitivity warnings.
Last edited:
I just found it, first off I would recommend choosing a much quieter Op-Amp for the phono stage than the 2134, something along the lines of the LM4562 would be a more appropriate choice.
I would boost the raw supply voltage to 30V if possible, if not it is certainly adequate for the application. (SLA batteries?)
Also in terms of cartridge loading there are a few cartridges that like loading at 100K, so options I would choose would be 100K, 47K, 22K and perhaps 1K as certain HO MC like lower loading. I'd probably provide an option for a simple socket for any bogey value you might need to stick in there in the future perhaps in place of the 2.2M which isn't needed as long as you don't switch loads with no cartridge connected. (In which case you would switch off the power anyway.)
Possibly depending on the op-amp you select you might consider inserting a very high quality cap in series with the gain setting switch to force the dc gain to unity reducing the amplification of input offsets - which in any event are not likely to be a big problem with the LM4562 (10nA input current bipolar) or the 2134 which is has jfets at the input. (IMHO it is still too NOISY..) The LM4562 also has the virtue of sounding very good and having SOTA performance levels for an audio op-amp.. It also has way more than sufficient open loop gain to maintain good RIAA accuracy at all frequencies of interest. It is a dual so only a single device is required.
http://www.national.com/pf/LM/LM4562.html#Overview
Possibly you might want to consider single op-amps to maximize separation as this will allow you to segregate the signal paths sufficiently to make capacitive coupling between traces a non issue.. The LME49710 is the device to choose..
http://www.national.com/pf/LM/LME49710.html#Overview
These are literally amongst the best performing audio op-amps you could ever hope for, and yes this tube-o-holic uses LM4562 in his modded Sony SCD-777ES SACD player.
The next thing I would do would be to scale the RIAA components by an order of magnitude lower to improve noise performance. Resistors down in value and capacitors up. This may be good for up to 6dB or more on the SNR front particularly in the region where your ears are most sensitive. A high impedance EQ network is not required in this application due to the output current available.
Finally I would add a unity gain buffer after the volume control, and given the current capability of the chips in question I would use a 10K volume pot for both noise and frequency response considerations. The BUF634 running open loop ought to do this pretty well I'd think.
On components, Wima and others make excellent and good sounding film caps. Resistors should be metal film for lowest noise. If using SMD resistors 1206 package are pretty easy to work with and either thick or thin film resistors should be fine in this application. (Use 1% tolerance - they don't cost significantly more) RIAA resistors should be 1% tolerance or better, 0.1% in SMD aren't expensive and the gain in accuracy is worth the extra pennies. Caps should be matched and tested to better than 1% as well. (Avoid SMD films as they are literally impossible to hand solder.) Note that 1206 size parts are easily soldered with a conventional solder iron.
I would boost the raw supply voltage to 30V if possible, if not it is certainly adequate for the application. (SLA batteries?)
Also in terms of cartridge loading there are a few cartridges that like loading at 100K, so options I would choose would be 100K, 47K, 22K and perhaps 1K as certain HO MC like lower loading. I'd probably provide an option for a simple socket for any bogey value you might need to stick in there in the future perhaps in place of the 2.2M which isn't needed as long as you don't switch loads with no cartridge connected. (In which case you would switch off the power anyway.)
Possibly depending on the op-amp you select you might consider inserting a very high quality cap in series with the gain setting switch to force the dc gain to unity reducing the amplification of input offsets - which in any event are not likely to be a big problem with the LM4562 (10nA input current bipolar) or the 2134 which is has jfets at the input. (IMHO it is still too NOISY..) The LM4562 also has the virtue of sounding very good and having SOTA performance levels for an audio op-amp.. It also has way more than sufficient open loop gain to maintain good RIAA accuracy at all frequencies of interest. It is a dual so only a single device is required.
http://www.national.com/pf/LM/LM4562.html#Overview
Possibly you might want to consider single op-amps to maximize separation as this will allow you to segregate the signal paths sufficiently to make capacitive coupling between traces a non issue.. The LME49710 is the device to choose..
http://www.national.com/pf/LM/LME49710.html#Overview
These are literally amongst the best performing audio op-amps you could ever hope for, and yes this tube-o-holic uses LM4562 in his modded Sony SCD-777ES SACD player.
The next thing I would do would be to scale the RIAA components by an order of magnitude lower to improve noise performance. Resistors down in value and capacitors up. This may be good for up to 6dB or more on the SNR front particularly in the region where your ears are most sensitive. A high impedance EQ network is not required in this application due to the output current available.
Finally I would add a unity gain buffer after the volume control, and given the current capability of the chips in question I would use a 10K volume pot for both noise and frequency response considerations. The BUF634 running open loop ought to do this pretty well I'd think.
On components, Wima and others make excellent and good sounding film caps. Resistors should be metal film for lowest noise. If using SMD resistors 1206 package are pretty easy to work with and either thick or thin film resistors should be fine in this application. (Use 1% tolerance - they don't cost significantly more) RIAA resistors should be 1% tolerance or better, 0.1% in SMD aren't expensive and the gain in accuracy is worth the extra pennies. Caps should be matched and tested to better than 1% as well. (Avoid SMD films as they are literally impossible to hand solder.) Note that 1206 size parts are easily soldered with a conventional solder iron.
Last edited:
Mag phono in
So, the actual installed mag phono input of my disco mixer looks a lot like his circuit. Does this mean I can increase the .001 and 0033 uf capacitors by 10 and decrease the 1.5m and 100k resistors by 10 and achieve less noise with a nice RIAA rolloff? Do I also have to decrease the 1k attached to plus input to maintain the same gain?Original circuit was for 4558 op amp, 33078 was my idea, and Newark's. I think 33078 can drive 600 ohm to some large voltage, whereas 4558 is limited to 2k load if I read it right.
So, the actual installed mag phono input of my disco mixer looks a lot like his circuit. Does this mean I can increase the .001 and 0033 uf capacitors by 10 and decrease the 1.5m and 100k resistors by 10 and achieve less noise with a nice RIAA rolloff? Do I also have to decrease the 1k attached to plus input to maintain the same gain?Original circuit was for 4558 op amp, 33078 was my idea, and Newark's. I think 33078 can drive 600 ohm to some large voltage, whereas 4558 is limited to 2k load if I read it right.
Last edited:
Yes you can actually scale all caps up by a factor of 10 and all resistors including that 1K down by a factor of 10. Note that the 22uF cap would have to increase to 220uF which could be a problem. (size and quality) It may be possible to dispense with this capacitor if input currents don't result in excessive dc offset at the output. One caveat I should mention is that the op-amp will be driving 100 ohms at very high frequencies. (Note this happens well above 100kHz, and the capacitive reactance at 20K of both caps in series is slightly over 1K ohm with the proposed values. It's still above 600 ohms overall at 40kHz including that 100 ohm resistor. )
The 33078 is a good op-amp, I used a lot of them in MI onboard pre-amps due to their good performance and low power consumption. The new National family I referenced represents a significant improvement sonically (noting IMO) and they are available in DIPs.. Quad version listed below.
LME49740 - Quad High Performance, High Fidelity Audio Operational Amplifier
They are very stable despite their excellent AC performance, good stability margin and are not generally very finicky in use. They just need to be have their rails properly decoupled relatively locally.
Last edited:
divide by 10 RC parts
Thanks for looking kevin. I was wondering if that 1.5M resistor was ideal- I know random noise goes up with the resistance. I just replaced the 25 uf capacitor with a +80-20, didn't realize it was part of the rolloff curve. So I will get 220 +-20%. Next month's pension check fun money is going to mouser, since they have some organ connectors I can't get anywhere else, and they never heard of an LM4562. Still haven't found a DIP LM4562 datasheet; datasheet catalog has only NJM 4562 which is a part with a lot less minimum specs than the NJM4560. I dont' know why anybody would buy NJM4562. Even the 15 MHz Gainbandwidth 33078, I had to put .1 at the dip socket +- and 20 pf across the feedback resistor to kill a 1mhz oscillation. But it sounds so much better than the 4558. .0033 and .001 cap on the disco mixer are ceramic with no percent listed, I can probably do better with plastic film caps if having them an inch or more away is not important.
Sorry for treading on your thread, Dano.
Thanks for looking kevin. I was wondering if that 1.5M resistor was ideal- I know random noise goes up with the resistance. I just replaced the 25 uf capacitor with a +80-20, didn't realize it was part of the rolloff curve. So I will get 220 +-20%. Next month's pension check fun money is going to mouser, since they have some organ connectors I can't get anywhere else, and they never heard of an LM4562. Still haven't found a DIP LM4562 datasheet; datasheet catalog has only NJM 4562 which is a part with a lot less minimum specs than the NJM4560. I dont' know why anybody would buy NJM4562. Even the 15 MHz Gainbandwidth 33078, I had to put .1 at the dip socket +- and 20 pf across the feedback resistor to kill a 1mhz oscillation. But it sounds so much better than the 4558. .0033 and .001 cap on the disco mixer are ceramic with no percent listed, I can probably do better with plastic film caps if having them an inch or more away is not important.
Sorry for treading on your thread, Dano.
Last edited:
The NJM part is something else.. Check link to National website for the LM4562 here: LM4562 - Dual High Performance, High Fidelity Audio Operational Amplifier, Note also that there is a quad version the LME49740. The LM4562NA (dual dip version) is in stock at digikey for $3.05 each, link here: Digi-Key - LM4562NA-ND (Manufacturer - LM4562NA/NOPB)
The lower feedback values should make the circuit much less prone to oscillation and IIRC the 4562/49740 devices have more phase margin than the 33078 which I recall having oscillation issues with.
Hope this helps.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.