DIYHiFi.org

For the sake of audio
It is currently Tue Aug 22, 2017 5:29 am

All times are UTC




Post new topic Reply to topic  [ 54 posts ]  Go to page Previous  1, 2, 3, 4  Next
Author Message
PostPosted: Sat Aug 05, 2017 1:14 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
hanski wrote:
I'm hoping that T. (after a couple beers) will come up with a raging transconductance sand hound that can be trained to a well behaving guard dog with degeneration.
:thumbsup:
Rant: it is a pita to import parts into brazil. Or should I say pitp: pain in the pocket! This country sucks in a number of levels.

At least in politics we are past the "FUBAR" point (we´ve been through it and may be initiating recovery. I certainly hope so.)


Top
 Profile  
Reply with quote  
PostPosted: Sat Aug 05, 2017 2:19 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
alexandre wrote:
This country sucks in a number of levels.

It very much does. I´m not particularly attached to the country, and I´m more italian than brazilian (citzenship and all), so leaving is always a possibility. I have lived and studied in Florence, the cradle of the renaissance, for two years (I was much younger - I´m 36 now). No girlfriend (and I like women only, very firm on that, thank you. I think it is appropriate to specify it these days.). No progeny. But I am attached to my parents and two sisters and we have a good living standard here, even though I´m not formally employed as of now. I am an EE.


Top
 Profile  
Reply with quote  
PostPosted: Sun Aug 06, 2017 1:09 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
alexandre wrote:
My desktop computer died earlier, while I was typing a post!

As an experiment I tried to resurrect the PC by soldering oscons under the suspicious cpu capacitors, and it worked! :thumbsup: It´s powering up again. Before this it would just spin the fan without displaying anything.

BTW the Vcore capacitors are all connected together, it´s one big bank of caps even though they are distributed around the cpu. They are 4V 680uF caps. And the oscons that I soldered underneath are 270uF.

I disconnected my HDDs of course. And I have backup of important data. This is just an experiment, the computer needs replacement.

Attachment:
mobo.jpg


You do not have the required permissions to view the files attached to this post.


Top
 Profile  
Reply with quote  
PostPosted: Sun Aug 06, 2017 5:40 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
It is now officialy an audiophile motherboard - six oscon SPs and four panasonic FMs :finga: One pana FM is not seen, it´s near the RAM. Too bad I don´t have any Lelons, my favorite, to make it sound good (just kidding!) :finga:

Seriously, it is working better than new. One processor that I thought was busted is working now: a core 2 duo E8600. This would not work even when the mobo was 'good'. So much for Intel's TDP spec (thermal design power) - it does not correlate with the electrical power needed for one particular processor. TDP is only a guideline for the family of processors. I have estabilished without a doubt that an E5400 processor needs less power and runs cooler even though TDP is specified the same 65W of the E8600. Obviously the cooler running E5400 will generate less disturbance on the power distribution and consequently the whole arrangement radiates less.

I have observed that the wireless usb adapter works better (almost doubled the throughput and halved the BER - bit error rate as reported by a manufacturer utility). I imagine that gaming motherboards, with their ridiculously expensive prices, might actually have something going for them. The EMI should be lower than that of an entry motherboard like this one which comes with lesser caps. Might have an extended life too.

Attachment:
audiophile mobo.jpg


You do not have the required permissions to view the files attached to this post.


Top
 Profile  
Reply with quote  
PostPosted: Sun Aug 06, 2017 6:56 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
alexandre wrote:
It is now officially an audiophile motherboard - six oscon SPs and four panasonic FMs :finga: One pana FM is not seen, it´s near the RAM. Too bad I don´t have any Lelons, my favorites, to make it sound good (just kidding!) :finga:

Actually, I wouldn´t be surprised if the RME pro soundcard sounded better now (through its analog outs). But I´m not too interested in the analog outs, I only use the digital ins and outs (toslink) with my own diy dacs which are clock master.


Top
 Profile  
Reply with quote  
PostPosted: Mon Aug 07, 2017 2:45 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
Here´s a way to make very good cap banks for your DIY projects. Cut a strip of double sided FR-4 PCB and solder the caps on the edge making sure the terminals are touching the copper on both sides. Hmm... I could have soldered a little closer to the cap body.

This one is an upgrade for the raw supply of my I/V converter. For the second 'C' in CRC. Most of my projects are single supply.

Have a nice week fellas.

Attachment:
IMG_20170806_232917980.jpg


You do not have the required permissions to view the files attached to this post.


Top
 Profile  
Reply with quote  
PostPosted: Tue Aug 08, 2017 3:42 am 
Offline
Muriel
User avatar

Joined: Thu Jul 26, 2007 5:45 pm
Posts: 1111
Hi,

hanski wrote:
So, is this the best circuit topology for a sand cell (leaving out cascoding and buffering)?
Anything left in the deep pockets?


Well, the one I use in products cascodes the J-Fet to keep Vgs at around 6V, using a trick circuit that biases the cascode with a zener in the drain of the second p-channel/pnp device. This circuit needs some trick to start up, but has benefits otherwise. One may also add a zener in the drain line of the J-Fet to cascode the p-channel/pnp device. Such cascoding tends to improve objective performance, especially HF distortion rise, with only increased dropout voltage as a problem.

If this "the best"? Who knows.

It sounds good, measures rather well, is simple and while it does use looped feedback the loop is tight, inherently stable and adjusting various parameters (e.g. 2nd stage degeneration, 1st stage Gm & current, device type in 1st & second stage) allows a wide range of results to be gained. So if all we want / need is a decent transconductance cell, we can do worse, a lot worse.

Other applications may benefit from different topologies.

One can (obviously) see how this topology may be extended by adding an N-Channel buffer in the tail of the P-Channel device (can be self biasing to provide a boot-strapped load to the P-Channel device) and then close the feedback from the source of the device to the input device source. We now have a classic 2-stage & buffer feedback circuit that parallels classic tube circuitry (Marantz Model 7 preamp style?) if we select gains and transconductance right. Further elaborations should be obvious to those "skilled in the arts".

Of course, we could just use an Op-Amp's (recent times have delivered excellent devices), but such a discrete circuit allows us to get every stage to behave "just so", trade many parameters freely and achieve results that may (or may not) sound more "interesting" or "better" than Op-Amp chip's where all those tradeoffs have been decided and fixed by others.

Ciao T

_________________
"For a successful technology, reality must take precedence over public relations, for nature cannot be fooled." Richard Feynman


Top
 Profile  
Reply with quote  
PostPosted: Wed Aug 09, 2017 6:34 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
Thanks T, my discrete circuits have not reached that level of elaboration but I´ll get there.

Take a look at the principle of my balanced DAC with TDA1543.

Attachment:
bias.png

I do not use the internal current bias (leave that pin disconnected for best results - ECDesigns hinted a while ago). The current sinks represent two TDA1543 chips pulling current to ground. Only the right channel of the chips is used because the left is flawed. The left channel picks up more interference from the switching of the inputs. Also, it does not latch simultaneously with the right channel (there is a one bitclock shift). "Balanced inside chip" would be a flawed approach with the 1543, as would be paralleling L&R.

(Has to be said, though, that I send the inverted data to the left channel even though I leave the pin disconnected. If you integrate the data line over time the net result is zero which may be beneficial for jitter.)

There are two situations depicted: idle and full swing. The bridge resistor is the active discrete I/V with its apparent impedance of 50 ohms (due to K170 Gm of about 40mS). Quite simple balanced BJT and JFET I/V, all N devices (not complementary). Since it´s a "bridge I/V" each chip sees 25 ohms // 1800, low enough for any DAC chips I choose (re-use of this I/V converter is desired). Notice the constant current on the 8V supply (this same supply powers the 1543 chips).

Virtually no gate current eliminates one error source that would be present in a common base I/V converter. Another benefit is there is only a twisted pair going from the stereo DAC board to each mono I/V board. No ground connection between DACs and IVs. Requires dual mono single supply I/V converters. And dual mono single supply floating amps. I can earth the dac board or the digital front-end if I desire. Digital interconnects are toslink (in and out) to an RME pro card. All is very stable, very RF immune (dead silent and no disturbance from fridge, lights, cellphone, nothing, even in close proximity. And sounds good! DC coupling without any servos is possible. In fact I have used it, but I prefer to place f-3dB at 20 Hz with my speakers.

The principle can be used with any DAC chips regardless of what DC voltage they expect. With AD1865, for example, I would place the two DC bias resistors to ground and use higher values since the idle is zero. I can tune and tweak this I/V and use it with any current output DACs changing the DAC board.

I might try TDA1545 again. I would use one chip per channel, balanced inside chip. It is different from TDA1543 in a number of ways and one of these is that the internal bias current cannot be disabled. If I use one NiMH cell for Vref (plus suitable bypassing) then I have 1.2V into the internal 11K resistor giving a reference current (Iref) of 109uA. Full scale current is then 109uA*13.2=1.44mA and midscale current is 720uA. Internal bias current is 109uA*9.42=1.027mA. The bias current (source) is added to the midscale current (sink) giving 307uA out of the chip at idle. This makes my DC bias resistors much larger and connected to ground. For 10K ohms resistors: 307uA*10K=3.07V (adequate). On the current out pin, internal to the chip, are mosfets sinking the bit currents to ground (there´s also a bit more circuitry such as the bias current source, this one comes from V+ since it´s a source current - I did take a decent look at the paper that describes the 1545).


You do not have the required permissions to view the files attached to this post.


Last edited by alexandre on Wed Aug 09, 2017 7:00 pm, edited 1 time in total.

Top
 Profile  
Reply with quote  
PostPosted: Wed Aug 09, 2017 6:52 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
alexandre wrote:
Here´s a way to make very good cap banks for your DIY projects.

Forgot to say, it needs a load right there on the double sided PCB. 4K7 works for me. It removes artificial brightness from the sound (probably an RF thing as the double sided board is a dipole antenna tuned to a certain frequency). A well tuned snubber might also be beneficial, I will try one.


Top
 Profile  
Reply with quote  
PostPosted: Thu Aug 10, 2017 3:42 pm 
Offline
Cow

Joined: Fri May 30, 2014 6:24 am
Posts: 122
alexandre wrote:
I do not use the internal current bias (leave that pin disconnected for best results - ECDesigns hinted a while ago). The current sinks represent two

I did browse through once the ECDesign's megathread at Diyaudio. It's long.. :D


Top
 Profile  
Reply with quote  
PostPosted: Thu Aug 10, 2017 4:00 pm 
Offline
Cow

Joined: Fri May 30, 2014 6:24 am
Posts: 122
Thanks T!
Kuei Yang Wang wrote:
Well, the one I use in products cascodes the J-Fet to keep Vgs at around 6V, using a trick circuit that biases the cascode with a zener in the drain of the second p-channel/pnp device. This circuit needs some trick to start up, but has benefits otherwise. One may also add a zener in the drain line of the J-Fet to cascode the p-channel/pnp device. Such cascoding tends to improve objective performance, especially HF distortion rise, with only increased dropout voltage as a problem.

Hopefully I got it correct:
(Unnecessarily high supply voltage, but I was testing the THD to 10k load -> 100V p-p output voltage).
Attachment:
Sziklai.png

I'm bending it for phono use (dreaming of needing just a a buffer amp after this) for my friend.

There would be open questions still:
1. Biasing the Sziklai. A 2MOhm resistor at J113 base and a servo (not yet in schema). Ok?

2.The 1kH inductor load should be replaced with an AC CCS or servoed CCS? (The 10k I/V resistor can't be used directly as Sziklai load because of high bias current.)

3. A good way to add tone controls? Another gain stage with tone controls, or something more clever?


You do not have the required permissions to view the files attached to this post.


Top
 Profile  
Reply with quote  
PostPosted: Fri Aug 11, 2017 5:30 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
I´m lost with this one, fellas. Never seen it. But this is good!
(You are clever Hanski, I wasn´t getting it from Thorsten´s post.)

It is a privilege to have you here helping us out, Thorsten. Thanks!


Top
 Profile  
Reply with quote  
PostPosted: Fri Aug 11, 2017 10:12 pm 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
alexandre wrote:
This one is an upgrade for the raw supply of my I/V converter. For the second 'C' in CRC.

That didn´t work out. I know what happened: they are RCRC filters and I ruined the Q (by adding a lot of caps at the wrong place). Now I made it Q=0.4 aproximately, with less capacitance, and it´s sounding good so far. Lesson learned.


Top
 Profile  
Reply with quote  
PostPosted: Sat Aug 12, 2017 2:23 am 
Offline
Goat

Joined: Mon Nov 28, 2005 1:10 pm
Posts: 514
There was something else wrong with my PSUs... better stop here to not embarass myself. Over to you fellas.


Top
 Profile  
Reply with quote  
PostPosted: Sat Aug 12, 2017 3:04 am 
Offline
Benjamin
User avatar

Joined: Mon Dec 13, 2004 3:29 pm
Posts: 4325
Location: Somewhere other than here.
hanski wrote:
I'm bending it for phono use (dreaming of needing just a a buffer amp after this) for my friend.


Many years ago, I wrote a program, in something called MathCad (late 80s), that showed what part to tweak, to fix RIAA inaccuracies.

What I did..................

We have what is a SOTA FFT (SOTA, for 1982!) that also has a synth function. We could enter in the parameters, for the RIAA network, and run that function. It was then stored, into memory, and then we could do a math function, to compare the two, and plot it in terms of relative deviation, from ideal.

With that, we could get +/- 0.02 dB accuracy!

So, having said all of that.......................

In a RIAA network, it is real easy to have a 0.25 dB "bump", in the 250-500 Hz region. It is worse, if it is in a feedback arrangement. Folks actually prefer that sound, so if you find that you have one turn out that way...................well, it isn't the end of the world.

Anyway, I do not know whatever happened to that program. If I did, I could post the "if you are too high (or low) here, you need to do x to y". Maybe I will look for it. No promises.


Top
 Profile  
Reply with quote  
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 54 posts ]  Go to page Previous  1, 2, 3, 4  Next

All times are UTC


Who is online

Users browsing this forum: No registered users and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group