|
|
Post by kcatthedog on Jun 10, 2017 19:35:29 GMT -6
Always impressed with the deep knowledge and experience that is only one question away here, you guys are gear head bad asses !!
|
|
|
|
Post by rowmat on Jun 10, 2017 19:41:22 GMT -6
Always impressed with the deep knowledge and experience that is only one question away here, you guys are gear head bad asses !! Is because they rack their brains!  |
|
|
|
Post by kcatthedog on Jun 10, 2017 19:58:35 GMT -6
Good one: oh that's why they a have those big ears !!
|
|
|
|
Post by rocinante on Jun 11, 2017 8:44:13 GMT -6
Lot of people reselling the 8 space api with compatibility issues.
Happy BAE user here.
|
|
|
|
Post by BradM on Jun 11, 2017 12:06:14 GMT -6
Given high frequency, it would be informative to have typical 20-20K bandpassed noise measurements along with full range (or 96K range) measurements, to see any difference in where various smps fall relative to one another, so it could be considered against any audible artifacts. Any more specific comparison of harmonic content outside of the normal non-distinguished THD %. Long day.....that's all I have for now..... Cool, thanks! I can do a 96k range. |
|
|
|
Post by BradM on Jun 11, 2017 12:06:52 GMT -6
Lot of people reselling the 8 space api with compatibility issues. Happy BAE user here. I'd love to hear more about this API issue you mention. Please elaborate. thanks, Brad |
|
|
|
Post by BradM on Jun 11, 2017 12:15:46 GMT -6
I meant the SMPS manufacturer in Taiwan. Neither API or Radial make them. Its just a catalog item for the manufacturer. I have no idea about the PSU for the different Neve racks. I have never seen the 6 sp LB. The one in the 10 space rack is an open frame medical type module. I did not dig in too far but could see it enough to recognize that. FWIW I do not think there was a noise dif when loading more modules in the 10 sp Neve rack but I have seen this with the brick style SMPS's. Its possible that Purple and Neve are using the same PS part in their 10 sp racks. The earlier style brick requires filtering. It was 20dB noisier than the new style that they are using. Filtering helped the old style by 10dB but does not make much dif on the new style at all. I think the internal filtering is much better. The other thing to check if you find the 3 rail brick SMPS is monitor the 48V rail while loading the modules one at a time. Do this with no load on the 48V and you will see it hit as high as 65V. It does drop rather quickly once loaded with 1 or 2 modules drawing phantom current. Yeah, that's what I meant. The supplier in Taiwan was providing the exact same unit to both API and Radial (this was a couple years ago...one of them might have changed parts/vendors since then). The reason I know this is because when I looked up the Radial part number, I found the datasheet with the label drawing showing the API graphics. Doh! Which earlier style brick are you referring to? And for which vendor? Who is the "they" you are talking about? Sorry you lost me with all the pronouns.  Interesting about the phantom supply. I really think that post-regulation after these SMPS line lumps is required. But it seems like not many are doing it. The Heritage OST10 looks like it might be good because it actually has individual regulators for each slot. Jeff - on your new 500 series rack you are working on...are you going to use the same power connector pinout as API and Radial? thanks, Brad |
|
ericn
Temp
Balance Engineer
Posts: 16,086 
|
Post by ericn on Jun 11, 2017 13:22:25 GMT -6
All I know is My Aphex lunch box has a Powerone module and my CAPI Rack has no problems running off a Little old DDA console supply trimmed down a wee bit!
|
|
|
|
Post by BradM on Jun 11, 2017 13:32:56 GMT -6
Sometimes the issues don't show up until you have circuitry that might be sensitive to noise on the rails. Do we assume that the modules must be designed for dirty power? Or do we assume that the power supplies should be noise free to a certain level? The problem is there is no spec for 500 series power supplies like the way there is for modules. There should be a universal ripple spec on the power supply rails. As a module designer, it's nearly impossible to determine all the interactions between supplies and other modules. For instance, I have an 1073LB here. When I turn the gain switch on that unit I can change the amount of noise I hear in modules a few slots over. That's kind of F-d up. jsteiger EmRR - What do you guys think about placing resistive loads on the power supply rails as a means of loading the power supply for testing. I'm thinking of making a little dummy module with some power resistors that I can stick into multiple slots. That way I'll have a repeatable, non-dynamic load. Thoughts? Brad |
|
|
|
Post by drbill on Jun 11, 2017 13:48:05 GMT -6
I meant the SMPS manufacturer in Taiwan. Neither API or Radial make them. Its just a catalog item for the manufacturer. I have no idea about the PSU for the different Neve racks. I have never seen the 6 sp LB. The one in the 10 space rack is an open frame medical type module. I did not dig in too far but could see it enough to recognize that. FWIW I do not think there was a noise dif when loading more modules in the 10 sp Neve rack but I have seen this with the brick style SMPS's. Its possible that Purple and Neve are using the same PS part in their 10 sp racks. The earlier style brick requires filtering. It was 20dB noisier than the new style that they are using. Filtering helped the old style by 10dB but does not make much dif on the new style at all. I think the internal filtering is much better. The other thing to check if you find the 3 rail brick SMPS is monitor the 48V rail while loading the modules one at a time. Do this with no load on the 48V and you will see it hit as high as 65V. It does drop rather quickly once loaded with 1 or 2 modules drawing phantom current. Yeah, that's what I meant. The supplier in Taiwan was providing the exact same unit to both API and Radial (this was a couple years ago...one of them might have changed parts/vendors since then). The reason I know this is because when I looked up the Radial part number, I found the datasheet with the label drawing showing the API graphics. Doh! Which earlier style brick are you referring to? And for which vendor? Who is the "they" you are talking about? Sorry you lost me with all the pronouns. Interesting about the phantom supply. I really think that post-regulation after these SMPS line lumps is required. But it seems like not many are doing it. The Heritage OST10 looks like it might be good because it actually has individual regulators for each slot. Jeff - on your new 500 series rack you are working on... are you going to use the same power connector pinout as API and Radial? thanks, Brad API and Radial have different pinouts as far as I know. |
|
|
|
Post by BradM on Jun 11, 2017 14:23:14 GMT -6
+1 I was actually thinking of grabbing one as I am attracted to the db25 connectivity for 8 modules - would interface well with my Redco patchbay and all of the reviews I've read on GS say it won shootouts among all the normal contenders. This, of course, doesn't probably include the random issues like Brad was taking about where 1 module would affect another 2 slots down. I will probably buy the CAPI if Jeff ever gets the new metal work for the new one and If suddenly learn how to solder (I try, and fail, over and over and have given up) Again this goes back to my comment about where does the responsbility lie for clean power - at the module or at the power supply? I think for most racks if the noise of a module is affected by another module, then that's probably an issue with the design of one or more modules. But it definitely gives people a bad taste in their mouth with regards to the rack being used since most folks don't have the the ability or the desire to determine where the fault lies. They just know that their gear is sucking. drbill - you are right...the Radial and API pinouts are different. I just purchased a Radial SixPack, by the way. Should arrive late this week. Brad |
|
|
|
Post by rocinante on Jun 11, 2017 14:23:57 GMT -6
Lot of people reselling the 8 space api with compatibility issues. Happy BAE user here. I'd love to hear more about this API issue you mention. Please elaborate. thanks, Brad Random 500 units werent working when fully loaded was what i read when i researched it. This was two years ago but people were sending them back. I wanted one too and almost went for it as others reported they worked great but i chickened out. This was on the purple site but iirc other places (online distributers) as well. |
|
|
|
Post by rocinante on Jun 11, 2017 14:26:00 GMT -6
I'll add that i owned a 6b for nearly a decade without a single issue and sold it for only a little less than i paid originally albeit i bought it used.
|
|
|
|
Post by BradM on Jun 11, 2017 14:26:33 GMT -6
I'd love to hear more about this API issue you mention. Please elaborate. thanks, Brad Random 500 units werent working when fully loaded was what i read when i researched it. This was two years ago but people were sending them back. I wanted one too and almost went for it as others reported they worked great but i chickened out. This was on the purple site but iirc other places (online distributers) as well. Is this before or after they went to the "high current" version they are currently offering? Brad |
|
|
|
Post by rocinante on Jun 11, 2017 14:35:14 GMT -6
Before. Its been awhile and i apologize for not doing my homework. So ignore me. A colleague was pissed when his didnt work and it obviously got to me but i guess they fixed the issue so once again forgive me.
*edit* i had a long anecdote here but it derailed from the topic.
|
|
|
|
Post by EmRR on Jun 11, 2017 14:43:33 GMT -6
jsteiger EmRR - What do you guys think about placing resistive loads on the power supply rails as a means of loading the power supply for testing. I'm thinking of making a little dummy module with some power resistors that I can stick into multiple slots. That way I'll have a repeatable, non-dynamic load. Thoughts? It would have value, not the whole story but at least give a quick and simple static view for direct snapshot comparison of PSU results. Are there a lot of 500 modules which have no onboard filtering at all? Or so minimal they don't take care of PSU/environmental contamination? Everything I can think of has diode isolation and fairly healthy filter caps. I suppose shared ground in the total environment may be the problem? If shared ground is the problem, what's the solution there? The only (non 500, but similar) module I can think of that lacks diode isolation is an RCA preamp which is essentially an API op amp type, it instead has inductance on the way in before filter caps. |
|
|
|
Post by BradM on Jun 11, 2017 14:56:34 GMT -6
jsteiger EmRR - What do you guys think about placing resistive loads on the power supply rails as a means of loading the power supply for testing. I'm thinking of making a little dummy module with some power resistors that I can stick into multiple slots. That way I'll have a repeatable, non-dynamic load. Thoughts? It would have value, not the whole story but at least give a quick and simple static view for direct snapshot comparison of PSU results. Are there a lot of 500 modules which have no onboard filtering at all? Or so minimal they don't take care of PSU/environmental contamination? Everything I can think of has diode isolation and fairly healthy filter caps. I suppose shared ground in the total environment may be the problem? If shared ground is the problem, what's the solution there? The only (non 500, but similar) module I can think of that lacks diode isolation is an RCA preamp which is essentially an API op amp type, it instead has inductance on the way in before filter caps. Some modules use resettable fuses or resistors along with the filter caps to provide decoupling (RC decoupling in both cases). The diode only offers protection against reverse voltage, no? My understanding is that it doesn't really isolate. Please educate me if I'm ignorant here. Ferrite beads are useful in decoupling high frequency noise that may come from switching supplies or digital circuits. I just started working on a little dummy load board. I have a power resistor from each rail to PGND. I also have an optional link for PGND to AGND. I then have test points for V+, V-, and PGND that will allow me to attach a 1X probe that will connect to my Prism dScope rig. Anything you guys can think of I may be missing? thanks, Brad |
|
|
|
Post by EmRR on Jun 11, 2017 15:37:18 GMT -6
Diode isolates in that it will prevent any power supply latch-up from unexpected directional draw at start-up. I've found cases with multiple supplies in which shared ground was a latch-up path if I didn't add diodes to the output of each supply. Little bit apples to oranges there. It's also rectifying whatever AC component exists before hitting module filtering. I can't say if that's better or worse than an inductive path, haven't thought through that. LDO's in the modules sounds like a reasonable approach if practical. |
|
|
|
Post by BradM on Jun 11, 2017 16:03:06 GMT -6
Diode isolates in that it will prevent any power supply latch-up from unexpected directional draw at start-up. I've found cases with multiple supplies in which shared ground was a latch-up path if I didn't add diodes to the output of each supply. Little bit apples to oranges there. It's also rectifying whatever AC component exists before hitting module filtering. I can't say if that's better or worse than an inductive path, haven't thought through that. LDO's in the modules sounds like a reasonable approach if practical. Interesting. LC filter would provide 12 dB/oct LPF. An RC filter will give you 6 dB/oct. With either of these decoupling schemes the trick is getting the cutoff frequency low enough to roll off low frequency noise like 60 Hz hum and its harmonics. Big caps are required, but that makes inrush current an issue. If you make the resistor bigger then you may drop too much voltage. It's always a tradeoff. I just did a quick test. I have less offending audible modulated noise in one of racks with an RC filter than I do with a diode/capacitor arrangement. I suspect a choke would be even better. Brad |
|
|
|
Post by Johnkenn on Jun 11, 2017 17:06:13 GMT -6
Lot of people reselling the 8 space api with compatibility issues. Happy BAE user here. I had major issues with my old Brent Averill... |
|
|
|
Post by rowmat on Jun 11, 2017 17:13:22 GMT -6
|
|
|
|
Post by BradM on Jun 11, 2017 17:18:48 GMT -6
Yeah, this one looks intriguing. I'm curious to know what regulators they are using on each slot. My hope is that they are using newer technology than the LM317/337, but even those will get the job done. There are some great sounding LDO regulators out there like the ones Little Labs is using in their Monotor headphone amp, for example. Brad |
|
|
|
Post by Bender on Jun 12, 2017 12:03:46 GMT -6
Jeff - on your new 500 series rack you are working on...are you going to use the same power connector pinout as API and Radial? Not to speak for Jeff, but I have one of the Proto's that was intended for Chris Stapleton's touring rig and I can confirm that the pin layout is EXACTLY the same as API. |
|
|
|
Post by jsteiger on Jun 12, 2017 18:01:20 GMT -6
Jeff - on your new 500 series rack you are working on...are you going to use the same power connector pinout as API and Radial? Not to speak for Jeff, but I have one of the Proto's that was intended for Chris Stapleton's touring rig and I can confirm that the pin layout is EXACTLY the same as API. I follow the same pinout as API. It is more or less their format so why other manufacturers like Radial, BAE, or OSA do their own thing is beyond me. |
|
|
|
Post by BradM on Jun 13, 2017 10:09:52 GMT -6
I totally agree. There needs to be an AES spec at this point. It's out of hand.
I just received some information from Purple about their supplies. The just started shiping the Sweet Ten with a new internal SMPS that has better noise performance and lower ripple, as well as improved fault protection features. It's also capable of supplying 2.4 A vs. 1.5A. I'm going to try to get one to include in my tests.
Brad
|
|
