• State Resolved
  • Locked Locked
  • Replies 22 replies
  • Subscribers 34 subscribers
  • Views 2643 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TAS5630 Over current protection problem.

Roger Wagg
Prodigy 50 points
Other Parts Discussed in Thread: TAS5630, TAS5615

We have used the TAS5630DKD in a design to drive a step-up transformer to produce 100V audio Line for use with commercial Public Address systems. Power output is to be 100W RMS based on a 27V DC supply.

Initially we used the evaluation PCB which easily acheived our design goal. However our protype PCB does not perform as well as the evaluation PCB.

We are using the TAS5630 in PBTL mode and it performs well upto about 85W then it appears to current limit and shuts down. To reset the amplifier the power has to be removed. The amplifier will operate continously at 80 Watts and the heatsink and inductors are running cool at 40deg C. Frequency response/output noise/ effeciency are all within spec.. The problem was repeatable over 5 prototype PCBs.

Current flowing into the primary of the transformer just prior to shutdown was 7.09amps RMS (peak to peak value 20 Amps). The evaluation board allowed a current of 10A RMS.

Changing the value of the OC_ADJ resistor made no significant change to the shutdown point. Components from the evaluation board were used in the prototype but there was no increase in output power before shut down occurred

Could anyone explain how the over current protection circuit works. What is its reference/sensing point.? The data sheet SLES220 (page 11)  states that the IOC in PBTM is infact 19A (I assume peak), but the evaluation PCB data sheets (SLAU287A Page 16) details 35Amps peak.

Any help or clarification would be very welcome.

  • 0
    TI__Expert 4320 points

    Hi Roger

    I have done some test with the TAS5630 loaded with a toroid step-up 100V transformer, and there was problems with saturation of the transformer partly due to the dc-offset of the TAS5630 and due to the low frequency signals fed to the transfomer.

    So when useing the TAS5630 you need to use a transformer that can handle the DC-offset - i.e. it needs some kind of air gap, so a El core or C-core, and only feed signals that don't saturate the transformer. Alternatively you need to apply a dc-servo to the TAS5630

     

    rgds,

    Kim N Madsen

  • 0 in reply to Kim Nordtorp Madsen
    Prodigy 50 points

    Hi Kim,

    Many thanks for the response.

    We were concerned about dc offset in the output transformer, so we incorporated a manual dc off set circuit . This was adjusted for zero offset as the output load  was increased. Unfortunately this made no difference whatsever.

    If I  understood how the current protection circuit worked I may be able to  find a solution.

    thanks again,

    Roger

     

  • 0 in reply to Roger Wagg
    TI__Mastermind 18510 points

    Roger, overcurrent protection in TAS5630 monitors current in each output device and starts cycle-by-cycle limiting (CBC) when current reaches the overcurrent threshold.  In CBC, each switching cycle is truncated at the point at which the overcurrent occurs, and the device does not shut down.  However, after about 5mS of CBC TAS5630 does shut down and asserts /SD.  The nominal overcurrent limit with Rocp = 22k or 24k is 19A, but tolerances and dynamic temperature effects can reduce it to 16A or so.

    However, I would like to echo what Kim Madsen said.  We have seen cases before in which the core of an output transformer saturates with a low frequency input, and this can certainly cause a device like TAS5630 to go into overcurrent and shut down.  This is an issue at higher output and not a DC offset issue.  There are several ways to deal with this, depending on the situation, and these are described in the attached applications paper SLOA133.  I hope this will help you.  Of course, the surest way to solve this kind of problem is to chose a transformer with a primary saturation frequency at full output that is well below the lowest audio frequency that will be used.

    Best regards,

    Steve.

    sloa133.pdf
  • 0 in reply to SteveCrump
    Prodigy 50 points

    Steve,

    Thanks for the document on transformer saturation. Very informative.!

    Our transformer was designed to work down to 50Hz and we have a 150Hz  high pass filter on the audio input so transformer saturation shouldn't be a problem. It is interesting that we had no problems with the same transformer on the evaluation board.

    Thanks for the description of CBC. This explains the CB3C block on the block diagram  Page 8 of SLES220. (Well, except for the 3.!!)

    You stated that the nominal overcurrent is nominally 19A can I assume that in PBTL this is doubled.

    How important is the need to ground the heatsink? The heatsink is black anodized aluminium and as such may not have a reliable ground connection.

    Thanks

    Roger

  • 0 in reply to Roger Wagg
    TI__Mastermind 18510 points

    Roger, grounding the heatsink is vital.  The heat slug conducts substrate currents which must be returned to ground.  The bottom surface of our heatsink is machined after anodizing to provide a good electrical as well as thermal contact.

    In general, it is very important to follow the rules about PCB layout shown in the data sheet and the EVM PCB layout.  If your grounding is not as good as that this could cause the problems you are seeing.  My advice is simply to copy the EVM PCB layout as closely as possible.  This could be the difference between your layout and the EVM.

    Regards,

    Steve.

  • 0 in reply to SteveCrump
    Prodigy 50 points

    Steve,

    Grounding the heatsink solved all our problems.! The difference in performance is amazing.

    Cannot thank you enough..

    regards

    Roger

  • 0 in reply to Roger Wagg
    Intellectual 600 points

    adding a resistor of a few ohms and bypass this with a 100u bipolar cap. this defines a reasonable DC resistance as  output load. The dc resistance of the transformer primary is much too small to drive directly from an amp with low output impedance.

     

    Geert

  • 0 in reply to geert de vries
    TI__Mastermind 18510 points

    Geert, I believe this is correct in some cases, but I do not think it is always true.  We have customers who drive transformer primaries directly and are successful.  This is possible when the following 2 conditions are met.

    - The DC current driven into the primary winding by amplifier DC offset is small enough that it does not bias the transformer core signficantly;  and

    - The additional quiescent current from the power supply that is produced by this DC current is acceptable.

    If these 2 conditions are not met it will be necessary to use a high-value DC blocking capacitor in series with the tranformer primary, or a circuit like the one you described.

    Regards,

    Steve.

     

  • 0 in reply to SteveCrump
    TI__Prodigy 65 points

    Steve,

    To you answer about heatsink, customer has follow-up question:

    There is no mention of this in datasheet tas5630.pdf except for this on page 25: Note T3: Heat sink must have a good connection to PCB ground.

     

    The heat sink on the EVM has a machined bottom surface but it is not available anywhere. The heat sink I intend(ed) to use is a quarter-brick heat sink (probably Thermalloy 241409B91200G http://www.aavidthermalloy.com/products/dctodc/dcdc_Quarter.pdf) with a 30mm fan nearby. It has an insulating ‘High performance QPad3TM interface pad’ on its bottom side. Should I be concerned? If so, what can I do about it? Must I machine the bottom of the heat sink?

    Thanks for your help.

  • 0 in reply to Prachi Patel
    TI__Mastermind 18510 points

    Prachi, as I mentioned above, grounding the heatsink is vital.  It is also vital to give good electrical and thermal contact between the heat slug and the heatsink.  (Data sheet SLES220B is clear about the thermal contact, which is mentioned 2 places, but not so clear about the electrical contact.)  The heat slug conducts substrate currents that must be returned to ground.  The bottom surface of our heatsink is machined after anodizing to provide good electrical and thermal conduction.  The customer will need to be sure of the same electrical and thermal contact, and it sounds like that will require machining the bottom of the target heatsink.

    Regards,

    Steve.

  • 0 in reply to SteveCrump
    Intellectual 685 points

    Steve, I am the customer that Prachi was referencing.

    I seen nothing in SLES220B (tas5630.pdf) which says the package heat slug must be grounded. Please tell me where to look for it.

    Sheet 1 of the datasheet says "Achievable output power levels are dependent on the thermal configuration of the target application. A high-performance thermal interface material between the exposed package heat slug and the heat sink should be used to achieve high output power levels."  Most thermal greases are electrically nonconductive.  What is a recommended electrically conductive thermal interface material?

    The heatsink used by the EVM apparently is unavailable anywhere. Machining a heatsink or making a custom heatsink will turn a limited-production item with reasonable cost into a much higher-cost item. Are there any alternatives you can suggest?

     

  • 0 in reply to Russell May
    TI__Mastermind 18510 points

    Russell, you are correct that the data sheet is not explicit about the electrical connection.  That is an oversight in its writing.  That's what I meant above.

    We use a product called Arctic Silver Ceramique that provides good electrical and thermal contact.  It is available at Radio Shack.

    Thermal contact is achieved pretty easily, even with an anodized or otherwise insulated surface.  It's necessary to add the electrical contact.  I don't know what the simplest way to do that is.  We have always chosen the secondary machining step to avoid adding any other grounding component.

    I am working on a source for heatsinks for ICs like TAS5630, but that program is still in startup.  I am sorry I can't offer that to you now.

    Regards,

    Steve.

  • 0 in reply to SteveCrump
    Intellectual 600 points

    Hi,

     

    Concerning the heat slug grounding of the TAS bridges we found the following solution.; we put a small strip of 0.1mm thick copper foil betwee the heatslug of the chip and the thermal interface material we use. The foil is soldered to the PCB-groundplave at one edge. The thermal interface material we use on top of that is Sarcon GR ( Fuji ) which is 0.5mm thick and allows for some mechanical tolerance between the amplifier board and the heatsink. This works well regarding heatsinking and grounding.

     

    regards

    Geert de Vries

    Duran Audio BV

  • 0 in reply to SteveCrump
    Intellectual 685 points

    From http://www.arcticsilver.com/ceramique.htm :

    Céramique does not contain any metal or other electrically conductive materials. It is a pure electrical insulator, neither electrically conductive nor capacitive.

    That implies some other electrical connection is necessary, such as the grounded foil suggested by Geert de Vries. About how much current must it conduct? About how much electrical conductance is appropriate?

    Some diamond-based thermal materials have even lower thermal resistance than Céramique. The pad attached to the bottom of the quarter-brick heat sink is probably worse. It might, or might not, be sufficient.

    Russ

  • 0 in reply to Russell May
    TI__Mastermind 18510 points

    Russell, Ceramique may not be highly conductive or capacitive, but in the extremely thin layer that is used between the IC and heatsink there is enough capacitance and conductance to ground the heatsink adequately.  In cases where it is not possible to make the area of the heatsink that contacts the IC heat slug conductive, it is possible to use a solution like Geert de Vries's.

    We have found it effective in our EVMs as well as in numerous cases in which users have insulated the heatsink from the IC without realizing the problem.  (Roger Wagg's case is one of these.  Problems he had were solved by providing the required electrical contact.)

    Regards,

    Steve.

  • 0 in reply to SteveCrump
    Intellectual 685 points

    Steve, I don't have a TAS5630 EVM but I do have three TAS5615DKD EVM's. We removed the heat sink from two of them. The thermal compound is a very good electrical insulator. On the oldest EVM, the anodized heat sink is machined so electrical contact between the heat sink and the IC's thermal tab is possible by accident or high pressure. On a later EVM, the anodized heat sink is machined only around the bolt holes so there is no electrical contact likely between the heat sink and the IC's thermal tab.

     

  • 0 in reply to Russell May
    TI__Mastermind 18510 points

    Russell, there is probably some contact between a heat slug and a properly machined heatsink, and there is clearly capacitive coupling at least, and these are necessary couplings for stability and performance.  This is demonstrated when people who have insulated the heatsink somehow follow our recommendation and ground it and connect it to the heatslug (whether with thermal compound like Ceramique or some other means).  Then problems caused by the insulation are corrected and the device performs correctly, a dramatic change.  So we will continue to recommend grounding the heatsink and connecting it to the heat slug with appropriate thermal compound.

    I think you are saying about the second heatsink that the surface that contacts the heat slug is still anodized, is that correct?  Then it is insulated electrically, and not by thermal compound (and this seems to indicate the dominant coupling is capacitive).  The anodized contact to the heat slug is a rather upsetting error.  I will try to arrange a replacement part and ask Prachi Patel to deliver it to you.

    Regards,

    Steve.

  • 0 in reply to SteveCrump
    Intellectual 685 points

    Yes, the surface that contacts the heat slug is still anodized for the second  EVM and probably the third EVM, because they arrived at about the same time.

    Prachi departed TI last Friday.

    Russ

     

  • 0 in reply to Russell May
    TI__Guru**** 189401 points

    Hi, Russell,

    Sadly, we have scrapped all our TAS5615 EVMs since this part is Not Recommended for New Designs.

    I can replace these three EVMs with a different part number, OR you can ship them to me, and I can get the anodization machined off.

    Let me know which way you prefer to go.

    -d2

  • 0 in reply to Don Dapkus
    Intellectual 685 points

    I would rather have a TAS5630DKD EVM. We killed both of the TAS5615's recently doing experiments. The older EVM had been repaired twice before, so we might not repair it. We could use its heat sink to replace the heat sink in the second EVM if necessary.

     We intend to replace the TAS5615DKD in the second EVM with a TAS5630DKD. I have been told that should be an acceptable substitution despite having a couple of inductor and capacitor values different from those in the TAS5630DKD EVM. We are not intending to push it past about 160 volt-amps in PBTL mode.

    Our application is not audio so it might be a misapplication. I might open some more topics after we have done some more investigation. One topic is already open which is not related to the failures: 'TAS5630 offset voltage'. We intend to drive a mostly capacitive load via a step-up transformer. Resistance of the transformer primary winding is low, so I expect idling current to be significantly affected by offset voltage. If necessary, I can add some resistance in series with the transformer primary, but I would like to minimize this resistance.

     

     

  • 0 in reply to Russell May
    TI__Guru**** 189401 points

    Russell,

    I'm going to send two TAS5630DKD EVMs to Lance today. They should arrive in a few days, and then he can get them in your hands.

    -d2

  • 0 in reply to Don Dapkus
    Intellectual 685 points

    Thank you for the EVM's! We are using them now.

    The Q3 thermal interface pad, attached to the Thermalloy heat sink I originally intended to use, is slightly electrically conductive. The manufacturer (Bergquist) says Q3 is electrically conductive. I verified that by laying a couple of flat washers on the pad and measuring electrical resistance between them. Resistance was 200 to 500 ohms (2 to 5 ms conductance) depending upon how far apart were the washers. There has been no answer to how much current must be conducted nor what conductance is needed between the thermal slug and the heat sink. Is this small conductance sufficient?

    Russ