AD797 Phono Stage Build and Help Desk Thread
- Thread starter HypnoToad
- Start date
tabarddn51
Super Member
Hi Guy's, I have just changed over my tone arm from a 10.5 development arm (been listening to this for the last few months) to my normal 12" Temmad Merlin Titanium arm with VHD wiring. I cannot believe the bass reproduction I am now hearing, it is such a pronounced increase that I have had to turn my active base amp down, All I can say is Whow.
Sorry, but I tried to upload the .asc for the active and PASSIVE RIAA equivalent stages but apparently I can't as it has an invalid extension.
Does anyone know how to get round the limitation?
Also. to use the .asc you'll need to have .lib files for the AD797 and the LME49710 in your model directories. Plus, for some reason the AD797 symbol does not include the decomp pin. I tried adding the pin and it resulted in a discrepancy.
The AD797 model is correct- I checked open loop parameters with the too-few-pins symbol dropped in a test schematic, just to be sure, and the .lib file viewed using a text reader, has all the pins, so it's a mystery.
Talking about the passive RIAA- here's a schematic and the simulated result vs. ideal (together with the active RIAA result as a comparison).
It can be built on the AD797MCPRE board as the network is used to replace the 390 ohm interstage R.
Last edited:
Here's a list of component changes. I'll identify the RIGHT CHANNEL ONLY for clarity. I'll also make it so that the original adjustable Rs remain.Hi Wyn and welcome to AK, as I have said previously the AD797 phono stage started out as a cheap easy to build MC phono stage and has evolved a bit from there, the RIAA network values were used because they were easy to source and worked well, I did incorporate a high pass filter in an earlier design to account for the rising response but to my old ears couldn't tell the difference. To condense what you have posted could you please list the value changes for the RIAA network and the offset mods for clarity or post a bigger schematic, my old eyes can't make out the values on the schematic you posted, there too small for me.
I'll also use the original board identifiers where possible (I hope).
ImpR: UNCHANGED (1k pot)
GainR: UNCHANGED (2k pot)
Both can be replaced by fixed Rs if desired.
R15: WAS; 47ohm, IS; 10 ohm. 1/4 W metal film 1%
The AD797 gain becomes 1+(GainR/10), so for a gain of 69 GainR becomes 680 ohms.
CAUTION!!!!
DO NOT drop GainR below 300 ohms- or a gain of 31.
If gains less than 31 are required, say for a MC cartridge of greater than 0.5mv output @ 1kHz, 5cm/sec, then increase R15.
For example, R15 = 30 ohms and GainR=300 ohms would be fine for a stage gain of about 11 and a cartridge output of about 1.6mv @1kHz, 5cm/sec such as the Denon Dl-110. The Dl-100 cartridge has a winding resistance of 160 ohms.
Note also that high output MC cartridges are designed to operate into a 47k load, just like a MM and generally have a high winding resistance, so ImpR would have to be replaced.
In essence the amp becomes a MM input stage.
For cartridges with internal Rs greater than 1kohm or so the AD797 is not a good choice due to the large base current noise of the input devices.
CroR9 is added as a shunt from the NI terminal of OA3 to ground.
3300pF 2.5% or better polypropylene film.
R5a is added in parallel with R5. 4.7Mohms. 1/4 W metal film 1%.
C23A is added in parallel with C23. 1.8nF 2.5% or better polypropylene film.
The passive RIAA can be produced from the modified or original design by removing the RIAA capacitors (C23, C21 etc.) and replacing the interstage resistor or resistor cap combo (R11 etc.) with the following network.
to the left is the output of OA1, to the right the NI input of OA3.
All Rs are 1% 1/4 W metal film, all Cs are polypropylene film, 2.5% or better.
R29=22k
R23=100r
R17=91k
R28=3300r
C20=100nF
Cr0R4=680pF
Cr0R2=33nF
Last edited:
That seems to do it! Thanks.
Generally for the RIAA I either use the Laplace function call- which is simple as the pole-zero constellation is defined by the standard- or do as I have in this case- use a known inverse network to compare it against as I planned to use the known network as the basis for a passive RIAA implementation.
The AD797.asy.txt file needs to have the .txt extension removed and then placed in the LTspice \lib\sym\opamps folder.
The AD797.txt needs to be changed to a .sub extension and placed in the LTspice \lib\sub folder.
A .lib spice directive needs to be added for AD797.sub
You can replace the LME49710 with the equivalent opamp of your choice, for example the LT1115, also an RIAA amp, which is present in the ltspicexvii libs and saves a lot of messing about.
By the way, I believe that your RIAA circuit has the extra "Neumann lathe" pole which creates a difference at c.10kHz and above- specifically c. 0.6dB at 20kHz.
I checked it against my ideal network and then I modified the ideal Laplace RIAA pre-emphasis source that I use, adding the additional 3.18us TC, just to be sure.
I do not intend to correct for this due to its non-standard nature and the fact that the HF roll off provided to the lathes is neither standardized nor does it correspond to a single pole. Also it is often corrected for during the mastering process so a "corrective" TC would only accentuate the extreme HF beyond the intent of the mastering engineer. Plus anything that accentuates the ultrasonic peaking of MCs can, in my opinion, only be bad.
Below I show the ideal RIAA that I design to and my reference RIAA deemphasis conforms to.
Last edited:
Wyn, I just wanted to say that it's awesome to have you here at AK. As a former ADI employee, I know you by reputation so it's pretty cool to find you here contributing to this project.
All,
I have a new TT on the way, and have been in phono preamp research mode, so this project looks very interesting. Looks like I have a lot of reading to do to come up to speed...
-Gary
All,
I have a new TT on the way, and have been in phono preamp research mode, so this project looks very interesting. Looks like I have a lot of reading to do to come up to speed...
-Gary
Mine will probably stay as original. I remember doing a modified PAS well beyond obsession to the point I wish I'd first assembled it with zippers instead of hardware. This is no doubt better and I definitely appreciate the dedication to make it as perfect as possible.
Thanks Gary. It's always good to hear from an ex ADI guy.
I have several reasons for getting involved in this project, but mainly it's because I want to build a multi purpose RIAA/NAB/IEC preamp that can be used to replace the 26dB SUT +46dB MM input tube amp stage on my venerable and greatly tweaked Hovland HP-100 preamp and by making some minor changes also replace the playback electronics on my Otari MX-50 tape deck and, possibly, the playback electronics on a friends Studer A810 tape deck.
Why do I want to do this? Well, for the tape decks the answer is simple- they both use 1970s/early 80s era opamps (i.e. NE5532) with low unity gain BW and relatively high noise/distortion and I want to see what happens with improved electronics.
My friend went to the effort of replacing the caps on his Studer- costing several thousand dollars in the process- but it still uses antique active devices, so how much better could a full on amp upgrade be. Yes, I could go to a discrete FET/bipolar or a tube design- and that will likely happen, but first let's try the easy path.
As far as the phono stage is concerned. The SUT is custom, with higher winding inductances than normal to improve the LF response and silver wire to reduce the winding resistance a tad so I could use smaller gauge wire to drop the capacitance down a smidgin, and an extra Mu metal shield to reduce the pickup, but it still shares the same problems as every other SUT because of the high capacitance load that it necessarily presents to the cartridge, so I wanted to build an active input stage.
I'm very fond indeed of the Miyajima cartridges, and they have been my transducer of choice for a number of years, from a time even before they were available in the US, but in all honesty they don't measure well above 10kHz or so, as is easily verified. All of them, like all other cartridges, have a mechanical cantilever/tip/suspension resonance which is independent of the electrical loading and which cannot be corrected by reducing the load R and placing a shunt cap across the output, so this is absolutely not just a RLC network issue. The Shilabe has a "dip" of >1dB in the 4k-10kHz region then a several dB boost at c.20kHz, the Kansui has a reduced dip (c.0.6dB), and a reduced boost, the Madake seems to have no dip, but a +4dB or so peak at about 20kHz- with each sounding better/more "real" than the previous one.
All of these characteristics look like a LC resonance together with an additional pole, with the additional pole being moved further out in frequency for the Kansui (together with increased damping on the resonance) and far enough out to flatten the c. 10kHz response on the Madake.
The differences are quite audible. The Shilabe gives recessed voices and a more "distant" image, together with an impressive attack and a "tizz" on cymbals. The Kansui moves the voices forward, but reduces the "tizz" so the sound seems less dynamic than the Shilabe- better for classical than rock. The Madake seems to do it all equally well, while retaining the thick, full, real, Miyajima sound.
The first two use a large diameter hollow aluminum cantilever, the Madake uses the composite bamboo (at the suspension)/aluminum (at the stylus tip end).
Miyajima labs recommend an "optimal" load (for example c. 200 ohms//c. 0.68uF) but the optimal load that is impled from the use of their own SUT is different from the recommended one (60 ohms NO extra cap), and far different from what some reviewers claim is best (47kohm!!!).
So, the bottom line is I want to play with the loading with control that I can't get using a SUT/tube preamp so I need a precise RIAA characteristic to start with and 72dB or so of gain, and if it sounds good also, well, all the better.
Oh, and I also want to compare an active vs. a passive RIAA stage with equivalent accuracy.
Both the passive and active designs are within +/- 0.1dB 20-20kHz both simulated and measured.
Last edited:
I just want to take this occasion to warn people, once more, about the dangers of purchasing AD797s from Ebay.
I have a few AD797s courtesy of ADI and I also purchased a few from DigiKey. I thought that I'd check out some of the "cheap" ones available on Ebay, and I received some which I then put in the preamp and tested.
Clearly they were fakes- and there are two easy ways to tell.
1. The standing current was wrong (4mA rather than 8mA)
2. the gain-bandwidth product was wrong.
The current was easy to check- we have a number of convenient 10 ohm resistors in series with the supplies.
GBWP was also easy to test - I just input a c. 1mv p-p sine wave signal into the preamp and.looked for the -3dB point of the input stage (gain of 83 in this case) and it was about 11kHz- which meant the GBWP was c. 1MHz- two orders of magnitude lower than the AD797 (110MHz). There were other issues also- output offset was too high, the frequency response of the RIAA stage was way off at HF, and the noise was too high, but these are harder to measure, or even to assess by listening- surprisingly being 3dB down at 12kHz is not as immediately obvious as you might think.
I replaced the fake AD797 with a real one and I measured a -3dB point of c. 1MHz!
I replaced the good AD797 with an LME47910 and the result was c. 600kHz- thus proving that the LME was, in fact, real as it supposedly has a GBWP of 55MHz.
I wonder how many people have bought these fake devices and not even noticed. Certainly the feedback for the vendor did not include a single mention of fake AD797s even though hundreds were sold.
So, if anyone knows of a bipolar input opamp with moderate offset performance, 4mA current consumption and a GBWP of c. 1MHz I'd be interested to find out what I bought.
By the way, I have reported the vendor to Ebay and sent a message to the perpetrator informing him of his illegal actions. He, as you might expect, is located in China- Hong Kong to be precise.
I have a few AD797s courtesy of ADI and I also purchased a few from DigiKey. I thought that I'd check out some of the "cheap" ones available on Ebay, and I received some which I then put in the preamp and tested.
Clearly they were fakes- and there are two easy ways to tell.
1. The standing current was wrong (4mA rather than 8mA)
2. the gain-bandwidth product was wrong.
The current was easy to check- we have a number of convenient 10 ohm resistors in series with the supplies.
GBWP was also easy to test - I just input a c. 1mv p-p sine wave signal into the preamp and.looked for the -3dB point of the input stage (gain of 83 in this case) and it was about 11kHz- which meant the GBWP was c. 1MHz- two orders of magnitude lower than the AD797 (110MHz). There were other issues also- output offset was too high, the frequency response of the RIAA stage was way off at HF, and the noise was too high, but these are harder to measure, or even to assess by listening- surprisingly being 3dB down at 12kHz is not as immediately obvious as you might think.
I replaced the fake AD797 with a real one and I measured a -3dB point of c. 1MHz!
I replaced the good AD797 with an LME47910 and the result was c. 600kHz- thus proving that the LME was, in fact, real as it supposedly has a GBWP of 55MHz.
I wonder how many people have bought these fake devices and not even noticed. Certainly the feedback for the vendor did not include a single mention of fake AD797s even though hundreds were sold.
So, if anyone knows of a bipolar input opamp with moderate offset performance, 4mA current consumption and a GBWP of c. 1MHz I'd be interested to find out what I bought.
By the way, I have reported the vendor to Ebay and sent a message to the perpetrator informing him of his illegal actions. He, as you might expect, is located in China- Hong Kong to be precise.
Last edited:
tabarddn51
Super Member
Hi Wyn, Well we all suspected that the Chinese sourced AD797's were fake, although they measure incorrectly, they still do work in the phono amp & the phono amp still sound sensational. (well to my ears at least) So doing an A / B listening test between the original op-amp & the fakes can you hear any difference.
Cheers Derek
Cheers Derek
The answer is, yes, I can tell the difference, but it's not as immediately obvious as I thought it would be based on the measurement differences. The background is not as quiet or dark. The 'tizz" on cymbals is less real and metallic. Details are fuzzier- things sound, in general, a bit harsher, a bit "less real". It's along the lines of using polyester film caps rather than ,say, polypropylene. (although the Russian PETP (polyester) 2.2u caps are far less obviously different to me than the 2.2u polypropylene film caps that I also use, and conventional wisdom says that they should be inferior, or at least less predictable, so who am I to say?).
One thing, the differences will be reduced if the input stage gain is reduced. For example, a gain of 21 (47ohms to ground, 1k ohms FB) in the first stage will push the bandwidth up to 50KHz or so, reducing the response error from about 3dB at 10kHz to several tenths of a dB at 10kHz, and increasing the audio band feedback factor so at least the static distortion is reduced proportionally.
I do understand that most purchasers will just stick the ICs into the sockets and listen. Few will be as ridiculous as I am in actually measuring everything as well as listening.
Who knows, perhaps not all cheap Chinese sourced AD797s are actually fakes.
Wyn
Last edited:
tabarddn51
Super Member
Hi Wyn, Thanks for that, evaluation. Having worked in China for 8 years, while I developed & sold my Temaad arms. I can assure you they are all fake, but many seem to work well. I can remember one famous amp designer of listening to a clone of he design. He thought it sounded great & could not understand how they did it for the money. It's not hard to figure out as the typical worker will only earn between 1-25 & $2-00 / hour.
Cheers Derek
Cheers Derek
I was watching a video the other day of a Chinese worker putting together kitchen cabinet doors, he had hand made tools, no work bench just knelt on the concrete floor, was bare foot and used his feet to hold them while he assembled them, and put those things together quick smart.
Thanks!
I have heard from the vendor. He blames his supplier and has asked for me to return the parts. I refused and he refunded my money anyway.
I demanded that he stop selling the parts immediately, but they are still for sale as of this am so neither he nor Ebay have taken action.
By the way, I have been remiss in not saying what a great job you have done in designing these boards. You and Sachin are providing an invaluable service to the DIY community. Very well done indeed!Hi Wyn and welcome to AK, as I have said previously the AD797 phono stage started out as a cheap easy to build MC phono stage and has evolved a bit from there, the RIAA network values were used because they were easy to source and worked well, I did incorporate a high pass filter in an earlier design to account for the rising response but to my old ears couldn't tell the difference. To condense what you have posted could you please list the value changes for the RIAA network and the offset mods for clarity or post a bigger schematic, my old eyes can't make out the values on the schematic you posted, there too small for me.
