What circuit changes did you make to get that much gain? I have just been fixing up an early Quad 33 with newer components. Everything works fine, but I wish I could use my Denon 103. A schematic or just the resistor numbers with new values would be greatly appreciated.
3rd edition of The Art Of Electronics by Horowitz and Hill, contains a huge table which lists rbb' (and Va!) for quite a large number of BJTs.
The lowest values in the table are non obsolete, thru-hole devices that you can purchase today (12 Feb 2016) from multiple sources. You give money, they give parts. I know this to be true as I just bought 100 pieces each of PNP and NPN, from Digikey of all places. These transistors have lower rbb' than the famous, and famously out-of-production, 2SB737.
According to AoE 3rd edition (p.501),
The lowest values in the table are non obsolete, thru-hole devices that you can purchase today (12 Feb 2016) from multiple sources. You give money, they give parts. I know this to be true as I just bought 100 pieces each of PNP and NPN, from Digikey of all places. These transistors have lower rbb' than the famous, and famously out-of-production, 2SB737.
According to AoE 3rd edition (p.501),
- ZTX851 NPN has rbb' = 1.67 ohms and Va= 410V
- ZTX951 PNP has rbb' = 1.24 ohms and Va= 120V
- 2SB737 PNP has rbb' = 1.70 ohms and Va= 140V
Seems to be that 'switching transistor' is synonymous with 'thermally limited package'. Of course also an omission of an increasing amount of linear characteristic data.
This what I kept from internet advice. Looking carefully the upper resistor R1-5/6 is shorted when M1 high gain setting and has no DC function. No real need to change the upper resistor unless wanting to have M2 as was or more ideal. I suspect the 180R was added to the 220R on the PCB.
I have made a bespoke pre amp using Radio 2 input. It is better. The Quad is not too bad considering it's poor reputation. This seems to be not matching the 33 to the pickup and not totally the 33.
DL103 would need even more gain and is outside the scope of gain change, This change suits DL110. Radio 2 is rather good as it is 100 mV and high input impedance. A gain of 500 seems about right if using Radio 2 input. If the typical passive 75 uS and active 3180/318 uS a gain of 30 stage one and 17 ( 3180-318 uS ) @ 1 kHz stage two seems about right. A passive 2 uS on the output would help the Quad 33. 220R and 10 nF should work. It is fairly standard practice to use 220R at the output so a two in one circuit. In theory no output cap required as the 33 has a non polar input cap ( please check ) .
This is the sort of thing you will need. The first op amp just needs gain to complete. A gain of 17 is 24.6 dB stage two in the calcualtor values. For the first op amp standard op amp gain. Try low values as voltage swing is small. 1K and 33R come to mind. The lower arm capacitor might need to be 1000 uF if fitted ( Panasonic FC are OK, use the high voltage types as they measure better, or electrolytic non polar that also work well ) . MC33078/9 seems to work without one ( might sound better with ) . The second stage could use 2 K lower arm and 22 uF >100 V polyester.
http://www.crazy-audio.com/basics/phono-equalization/activepassive-riaa-online-calculator/
This is the sort of thing you will need. The first op amp just needs gain to complete. A gain of 17 is 24.6 dB stage two in the calcualtor values. For the first op amp standard op amp gain. Try low values as voltage swing is small. 1K and 33R come to mind. The lower arm capacitor might need to be 1000 uF if fitted ( Panasonic FC are OK, use the high voltage types as they measure better, or electrolytic non polar that also work well ) . MC33078/9 seems to work without one ( might sound better with ) . The second stage could use 2 K lower arm and 22 uF >100 V polyester.
http://www.crazy-audio.com/basics/phono-equalization/activepassive-riaa-online-calculator/
Last edited:
Thanks for that, both Nigel and rsavas. So if I am understanding this correctly, you can alter the value of R107/108 while keeping the sum of 105+107 and 106+108 constant not to disturb the riaa EQ. I have seen that Dada are recommending a similar change in their modding instructions.
I have changed the transistors on the PU board for BC550C. What would happen if I made R107/108 even smaller, would I get more gain or would the transistor start distorting or oscillating? (Im a tube hobbyist usually, sorry if the answer is obvious)
I have a standalone tube stage for riaa playback. Its DIY based on the old RCA circuit using ECC83s (its the wrong forum but I can post it if anybody wants to see). But of course that is also totally unfit for a DL 103. So maybe I could build that opamp stage and connect to Radio 2.
How about a pure gain stage without any correction? Could I just use a couple of JFETs in front of the PU input, pretending to be an MM cart? (I have some 2sk170 in the box). The 33 has a rather weak single rail supply, so something simple would be best to try.
I have changed the transistors on the PU board for BC550C. What would happen if I made R107/108 even smaller, would I get more gain or would the transistor start distorting or oscillating? (Im a tube hobbyist usually, sorry if the answer is obvious)
I have a standalone tube stage for riaa playback. Its DIY based on the old RCA circuit using ECC83s (its the wrong forum but I can post it if anybody wants to see). But of course that is also totally unfit for a DL 103. So maybe I could build that opamp stage and connect to Radio 2.
How about a pure gain stage without any correction? Could I just use a couple of JFETs in front of the PU input, pretending to be an MM cart? (I have some 2sk170 in the box). The 33 has a rather weak single rail supply, so something simple would be best to try.
KSA992 is worth a look. I suspect it is a Japanese wafer made up by Fairchild. Not unlike 2SA920/872A. Useful where 120 V needed. See NEC 2SA992 for noise graphs.
KSA992FBU | Fairchild KSA992FBU PNP Bipolar Transistor, -50 mA, -120 V, 3-Pin TO-92 | Fairchild Semiconductor
Mods to 33 seem as you say. I think changing the R107/108 is enough ( add in paralell ). The 180R is what I added to 220R or about 100R if replacing a resistor. That would be about 1.2 mV instead of 2.5 mV. The circuit is less likely to oscillate at higher gain as negative feedback is less. RIAA might change a little as the loop gain is at the limit when a simple circuit like this ( softer ). I doubt it would be drastically different. As the transistors are very low noise I suspect even higher gain is possible. Even 47R in paralell could work. 60 dB hiss would be no worse than a good tape deck.
KSA992FBU | Fairchild KSA992FBU PNP Bipolar Transistor, -50 mA, -120 V, 3-Pin TO-92 | Fairchild Semiconductor
Mods to 33 seem as you say. I think changing the R107/108 is enough ( add in paralell ). The 180R is what I added to 220R or about 100R if replacing a resistor. That would be about 1.2 mV instead of 2.5 mV. The circuit is less likely to oscillate at higher gain as negative feedback is less. RIAA might change a little as the loop gain is at the limit when a simple circuit like this ( softer ). I doubt it would be drastically different. As the transistors are very low noise I suspect even higher gain is possible. Even 47R in paralell could work. 60 dB hiss would be no worse than a good tape deck.
Last edited:
The DL103 is 14 Ohms. I thought the rbb' fetish was for a couple of absurdly low output carts and ribbon mics. When rbb' = re/2 diminishing returns set in so you are talking about Ic's of 10mA or more to take full advantage.
BTW Phillips gave an AES paper decades ago on a geometric exercise to minimize rbb'. I ended up with two diff-pairs (never used them) designed specifically for the ultra low output Ortophon MC cart. They never became a product so possibly these are the only ones in existence.
Last edited:
That makes sense (I was thinking of the DL103B ?). We ran some prototype devices for an old RCA ribbon mic experiment at MIT as a favor to Dr. Jerry Letvin (famous for telling Timothy Leary that basically he was full of s**t in public) only to find that his circuit required an absurdly low Vcesat at high Ic. Their prototype used the SB797, etc. parts and did work.
The claim was that a virtual ground pre-amp would mechanically damp the ribbon. Maybe, but they really wanted to do the next great MC pre-amp with the same claim (even filed a patent) which I still disagree with. All ended up for naught anyway.
Thx Mark,
You always supply info that others have yet to figure out or at least share with us. When is your book coming out
Wonder what the smt versions of these parts are? I guess look for similar characteristics. assumed to be in a sot-223
Since Zetex now diodes inc does not publish rbb or noise specs/curves, how do you know from the data supplied if they are low noise devices, Vcesat,Ic? Is the only way to find out is actually doing a noise test, such as NEC does/did for a 2sa992 or other low noise specific parts.
You always supply info that others have yet to figure out or at least share with us. When is your book coming out
Wonder what the smt versions of these parts are? I guess look for similar characteristics. assumed to be in a sot-223
Since Zetex now diodes inc does not publish rbb or noise specs/curves, how do you know from the data supplied if they are low noise devices, Vcesat,Ic? Is the only way to find out is actually doing a noise test, such as NEC does/did for a 2sa992 or other low noise specific parts.
I think Dr.'s H & H looked at data sheets for devices for key indicators (one of which is very low Vcesat) to find devices that would likely have low rbb' and then actually measured them to confirm. These are power devices that need low rbb' for other reasons, virtually no one specs parts for low noise audio/instrumentation as primary applications anymore. The BF862 is another example of fortuitous coincidence. These particular Zetex devices are nice in that they are TO-92, there are some other very low rbb' power devices but they are in TO-220 or TO-3.
curious how to measure rbb'?
Gee I should look in my own computer somewhere here i think
http://www.synaesthesia.ca/LNschematics.html
yup he made a test jig
Gee I should look in my own computer somewhere here i think
http://www.synaesthesia.ca/LNschematics.html
yup he made a test jig
Last edited:
Not easy at all a these low numbers. You have to de-embed it's effects on measurements of device characteristics, which is usually very inaccurate. Like weighing a pound of meat with an elephant on the scale. Or infer it from noise measurements vs Ic assuming re and Re are the only other sources of noise. Re can be estimated from log conformance measurements. In the end this is a serious lab exercise and the 3 digit numbers given are "enthusiastic".
EDIT - Mr. OP's measurements would be rbb' + Re a nit maybe, since noise is stochastic lots of digits in it's measurement implies a LONG time. In the end the output noise is the goal not quantifying rbb' in isolation.
Last edited:
When the little Sony /Ortonfon transformers were cheap ( T10 I think ) that was a better answer for many. Sony sold them trade for $10 a pair in the day. When dropped Ortofon sold them at 4 times the price which was still a bargain. When tested they were far better than looked possible.
The DL110 is 160 R and > 1mV. When a preamp is built for it the sound might rival DL103. Problem is neither MC or MM typically offered for it seem ideal. This causes people to think it nice, but nothing specail. Not true. DL110 seems to have ideal mass and compliance. It seldom is a mismatch with any PU arm.
The DL110 is 160 R and > 1mV. When a preamp is built for it the sound might rival DL103. Problem is neither MC or MM typically offered for it seem ideal. This causes people to think it nice, but nothing specail. Not true. DL110 seems to have ideal mass and compliance. It seldom is a mismatch with any PU arm.
Thanks for the ideas with the 33 preamp. I will be sure to try out what works in the next weeks.
Regarding FET input MC frontends, I found some simple circuits in this Tubecad article. The author warns about poor power supply noise performance, but as the Q 33 already has a well filtered and zener stabilized supply (uprated to 16v on mine) it could work. Or one could perhaps run the FET stage off its own 7812 regulator.
Link: Moving-Coil Pre-Preamplifiers
Regarding FET input MC frontends, I found some simple circuits in this Tubecad article. The author warns about poor power supply noise performance, but as the Q 33 already has a well filtered and zener stabilized supply (uprated to 16v on mine) it could work. Or one could perhaps run the FET stage off its own 7812 regulator.
Link: Moving-Coil Pre-Preamplifiers
I have some results when it comes to my DL103 into the MM input of the Quad 33. Of course I could not keep myself from cobbling together a test circuit just now. I can happily report that it works fine to put just a 2SK170 GR before the input. Im having no particular problem with noise or hum even though the circuit hangs in free air beside the preamp at the moment. I am using 100R for a cartridge load, and the 2 FETs are temporarily powered from the 16v line in the Quad through a 1k resistor + 220uf which drops the voltage to around 12v. Coupling to the next stage (MM input, "M1" setting with the gain change that Nigel Pearson suggested) happened to be through 0,68uf film caps that I had around.
I just thought others might like to know that it sounds fine this way, but can probably be tweaked for lowest distorsion and properly screened in a metal box. Thanks guys for providing inspiration.
I just thought others might like to know that it sounds fine this way, but can probably be tweaked for lowest distorsion and properly screened in a metal box. Thanks guys for providing inspiration.
Last edited: