I think joarcim is just testing the voltage headroom and the slew capability...
The paradise accepts rather high inputs...with a different RIAA circuit it could be configured for MM with 45dB gain as well... though distortion would suffer a bit...from app 90 dB to app 80 dB with the MM's higher input voltage-swing
The paradise accepts rather high inputs...with a different RIAA circuit it could be configured for MM with 45dB gain as well... though distortion would suffer a bit...from app 90 dB to app 80 dB with the MM's higher input voltage-swing
my first quick trial is nowhere near comprehensive, it was just a quick test, but the noise was very low, input DC offset (once the cartridge loading was connected) was not measurable, and soundwise.... well.... too early to tell. but I liked what I heard!
Ricardo, take the Lyra Titan i for example. It puts out 0,5mV @ 5cm-sec. 20cm-sec can be cut so we need 12dB overhead, better 20dB so 5mV can happen. I do not talk about miss tracking and ticks and pops. The Paradise is tolerant to miss behavior because of the nature of the topology. I have explained that before. 5mV into the Paradise is 5V out.
Up to 5V at the output the Paradise should hold RIAA precision. Maybe other phono stages have accurate RIAA at 1V output but what happens when the output is higher !
Yes, MiiB, my measurement shows the slewing behavior at higher level. I found that a stage that hold the shape of the square at higher level is better. The Paradise input stage is hard to slew because it is very fast. The limit is of how much current the mirror can drop into the RIAA. This gets better when the idle current in the input stage is high.
My original design had 5.5mA in each transistor with 20 Ohm degeneration. Now it is a bit over 2mA and 33 Ohm degeneration. I think this is more optimal for cartridges that have a bit higher impedance and a bit less output like the DL103.For my Titan i would prefer the original higher current. That may give more distortion though but it would slew even less because we have two times more current to dump into the RIAA.
Up to 5V at the output the Paradise should hold RIAA precision. Maybe other phono stages have accurate RIAA at 1V output but what happens when the output is higher !
Yes, MiiB, my measurement shows the slewing behavior at higher level. I found that a stage that hold the shape of the square at higher level is better. The Paradise input stage is hard to slew because it is very fast. The limit is of how much current the mirror can drop into the RIAA. This gets better when the idle current in the input stage is high.
My original design had 5.5mA in each transistor with 20 Ohm degeneration. Now it is a bit over 2mA and 33 Ohm degeneration. I think this is more optimal for cartridges that have a bit higher impedance and a bit less output like the DL103.For my Titan i would prefer the original higher current. That may give more distortion though but it would slew even less because we have two times more current to dump into the RIAA.
The reason I cut down the current is because of the dissipation on the cascodes....you can easily run more current. but you have to find the limit yourselves. by halving the current set resistor (R5=60 ohm) you double the current. but then you also have to reduce the 2X2 resistors (220 ohm) to 2x110 ohms.. Cascode dissipation will then be 150 mW... which to me is on the edge....for a small TO92 device.
Last edited:
Thanks Michael. http://www.ti.com/lit/an/slyt145/slyt145.pdf
Ricardo, you can grind your mathematical ax on circuit fig.16
It promises accurate RIAA.
Ricardo, you can grind your mathematical ax on circuit fig.16
It promises accurate RIAA.
I have talked to Jahre and i got some interesting answers. I wanted to know what the smallest case is with 1% precision for RIAA use. It is case 53.
Here are the dimensions :
6,8nF 125V 12,7 x 12,5 x 3,4 10.2mm between wires Case 53.10
13,6nF 125V 12,7 x 12.5 x 5,3 10.2mm between wires Case 53.20
40,6nF 125V 22 x 12.2 x 10,4 20mm between wires Case 53.50
i get a price quote soon
best, joachim
Here are the dimensions :
6,8nF 125V 12,7 x 12,5 x 3,4 10.2mm between wires Case 53.10
13,6nF 125V 12,7 x 12.5 x 5,3 10.2mm between wires Case 53.20
40,6nF 125V 22 x 12.2 x 10,4 20mm between wires Case 53.50
i get a price quote soon
best, joachim
Tanks Joackim.
It may take same time to get a proto board out but I have 4 days off from tomorow so we get there.
I have 100 m space on my ennclosure so I may go for that width for a one off Just to make shure.
There are loads of things to get finalised board size being just one.
Then Diodes orientation. and maybe TO 247 foot print
Space for heat sink if needed (thinking to use the same as on the Paradise)
and so on.
I will PM Frans with my email this will make possible to send out gerbers and DXF files
Tank you wery much for your kind words.
Input capacitance is 4 times that of the input transistor...BC337 is app 5 pF...So a good guess would be 20 pF... looking into low impedance on the source.. roll-of from this is nothing..
I will read it but it seems these use opamps and not an interstage passive filter...
Well is is actually 8 x , but anyway, BJTs have much less input capacitance ( and hence less open loop distortion ) then J-Fets so this is an advantage of BJTs.
Yes, Ricardo, it is with Opamps but that is not the point. They argue that a conventional filter has mathematical problems to follow the RIAA exactly and this arrangement of 4 Opamps solved that. I find it intellectual interesting, not more.
SK170 has input capacitance of 22 pF so this here with a total of 40 pF is roughly the same at one pair of fets, but here we have 4 sets of input devices with much much lower noise and way higher mu..
Last edited:
The P-channel 2SJ74 has even 105pF. That does really not matter much. An MC cartridge is so low in resistance that the pole created is very high in frequency.
Wima is a good option. I did talk to them some years ago to get special GTO caps for industrial project. They were very helpful and it was not a problem to buy small quantity in OK price.
FKP2 is (so far) copper in RM5 'bauform'. I can get easily 2.5% with no problem but 1% would be much better!!