Worked example: an open quadruped, Pupper and Doggo
Result. Two open quadrupeds have priced parts lists in this database, and they show two actuation choices. This site can show the prices of one of them.
- Stanford Pupper: the database holds a priced parts list from the Pupper sheet. No licence was found for that sheet, so this site shows none of its quantities, prices or totals. This chapter names its parts and gives the method only.
- Stanford Doggo: $2,971.02. Eight brushless motors with their drivers, encoders and belts are $1,313.36, 44.2 percent. That is $164.17 per motor.
- Pack prices are a trap. A BOM row can hold a pack price against a piece count, and a naive sum of quantity x price then overstates the cost. The Pupper sheet does this for much of its hardware. The Doggo sheet counts packs, so its sum is the buy cost.
- No torque check can be run on either robot from this database. It holds no torque, speed or mass for the servo or the motor, and no leg length for either robot. The papers' robot masses are in the database, but this site cannot show them. This chapter says what is missing and does not fill the gap.
The arithmetic is in tools/book_calcs/quadruped.py; its output is tools/book_calcs/out/quadruped.md. The Pupper part of that output uses the sheet's prices, which this site cannot show.
Pupper parts list
Source: the Stanford Pupper BOM sheet (pupper). The source record notes that no licence was found for the sheet. So this site shows none of its rows, and this chapter gives no Pupper cost. The parts the sheet names include JX CLS6336 hobby servos as joint actuators, a Raspberry Pi 4 for compute, and an optional IMU set (BNO080, Teensy LC). The sheet also prices printed parts and carbon fiber plates.
The pack trap
A catalog BOM row can hold a price for one thing and a quantity of another. On Pupper, many hardware rows hold a pack price against a piece count. The database moves those prices to price_note.
The pack rule: for an item in packs of N, buy ceil(total pieces over all its rows / N) packs. For example, if three rows ask for 40 screws in all and the screws come in packs of 100, buy one pack. A naive sum counts the pack price 40 times.
Rows that the sheet marks as repeats, and rows priced per length with a bare count, need a choice. State the choice next to the total.
The Doggo sheet avoids the trap. It names the pack in the item and counts packs, so quantity x price is the buy cost:
| Doggo row | Qty | Price each (USD) | Cost (USD) |
|---|---|---|---|
| M3x10 screws, 100 pack | 1 | 6.00 | 6.00 |
| Threaded insert pack, 100 pack | 1 | 11.89 | 11.89 |
| Dowel Pin, 10 pack | 1 | 5.79 | 5.79 |
| Outer Bearing, 2 pack | 4 | 8.98 | 35.92 |
Doggo parts list and cost
Source: the Stanford Doggo BOM sheet (stanford-doggo), marked work in progress in the source record. 43 rows, all priced in USD, 19 linked.
| Group | Rows | Sum qty x price (USD) | Share |
|---|---|---|---|
| Co-axial leg drive | 12 | 1,225.88 | 41.3% |
| Electronics | 15 | 889.73 | 29.9% |
| Legs | 6 | 411.40 | 13.8% |
| Body | 5 | 254.28 | 8.6% |
| 3D printed components | 1 | 163.30 | 5.5% |
| General hardware | 4 | 26.43 | 0.9% |
| Total | 43 | 2,971.02 |
Drive chain and custom parts, pulled out of those groups:
| Line | Qty | Unit USD | USD |
|---|---|---|---|
| T-Motor MN5212 KV340 | 8 | 82.92 | 663.36 |
| ODrive 24V | 4 | 119.00 | 476.00 |
| AS5047D encoder board | 8 | 15.75 | 126.00 |
| 70T GT2/3 timing belt | 8 | 6.00 | 48.00 |
| Drive chain | 1,313.36 (44.2%) | ||
| Waterjet links | 1 | 306.00 | 306.00 |
| Outer leg shaft | 1 | 247.40 | 247.40 |
| All 3D printed parts | 1 | 163.30 | 163.30 |
| CNC routed carbon fiber | 1 | 120.00 | 120.00 |
| Waterjet sheet metal braces | 1 | 105.30 | 105.30 |
| Custom-made | 942.00 (31.7%) |
Drive chain per motor: 1,313.36 / 8 = $164.17.
The two designs side by side
| Item | Pupper | Doggo |
|---|---|---|
| Joint actuator | JX CLS6336 | T-Motor MN5212 KV340, 8 |
| Actuators per leg | not shown | 2 |
| Catalog class | hobby servo | outrunner BLDC, with a timing belt stage |
| Reduction | inside the servo, per the class row; ratio not in database | 70T GT2/3 belts; ratio not in database |
| Driver | inside the servo, per the class row | ODrive 24V, one per two motors |
| Position sensor | inside the servo, per the class row | AS5047D, 8 |
| Controller | Raspberry Pi 4 | Teensy 3.5 |
| Battery | not shown | "Batteries", 2; no spec in database |
| Total | not shown | $2,971.02 |
| Custom-made share | not shown | 31.7% |
What the guidelines and class rows say about each choice. None of it is a measurement of these two robots, except where it names Doggo.
| Topic | Statement | Source |
|---|---|---|
| Low ratio | QDD uses a single-stage reduction under 10:1. It raises torque at the expense of control bandwidth and keeps the motor backdrivable | [1], from the Doggo paper |
| Motor scaling | With motor mass scaling with radius, torque scales with radius squared, reflected inertia with radius cubed | [2] |
| Outrunner heat | Outrunners trade thermal dissipation and inertia for torque density | [3] |
| Low ratio and heat | A low ratio can raise Joule heating for a given torque | [4] |
| Doggo's motor test | Raised current until the winding reached a steady 125 C | [5] |
| Doggo's commutation | Field-oriented commutation at 10 kHz; an example, not a limit | [6] |
| Hobby PWM servo class | No position, load or temperature feedback to the host | class row act-hobby-pwm-servo |
| Outrunner class | Rated for airflow in flight, so continuous current in a sealed joint is lower; datasheets are often headline figures only | class row motor-outrunner-bldc |
The class ranges below describe the outrunner family, not the MN5212. Do not read them as its values.
| Family | Attribute | n | Low | Median | High |
|---|---|---|---|---|---|
| Outrunner BLDC | KV | 102 | 65 rpm/V | 1,950 rpm/V | 27,000 rpm/V |
| Outrunner BLDC | Mass | 110 | 1.7 g | 33.8 g | 1.74 kg |
| Outrunner BLDC | Price | 80 | $10.90 | $25.90 | $689.90 |
The ranges count only the values this site can show. It can show a stall torque and a mass for only one hobby PWM servo, so the table gives no range for that family.
The $82.92 Doggo motor row is above the median outrunner price and far below the top of the range. The median mass of 33.8 g and median KV of 1,950 rpm/V show that at least half the family are small, fast motors, so the median price says little about a leg motor. The Pupper part's catalog class is hobby-servo; the database does not record whether it takes PWM.
The check you cannot run yet
The rated-torque check needs a load and a rated torque [7]. The method is the same as the knee check in chapter 20. Here is what it needs and what the database holds.
| Input | Pupper | Doggo |
|---|---|---|
| Robot mass | paper value, not shown; the control code models a 0.56 kg frame, four 0.08 kg leg modules and four 0.03 kg legs | paper value, not shown |
| Leg link lengths | not in database | not in database |
| Actuator rated torque | not in database | not in database |
| Actuator stall or peak torque | not in database | not in database |
| Reduction ratio | not in database | not in database |
| Bus voltage | not shown | 24 V in the driver's name only |
| Battery capacity | not shown | not in database |
| Upstream simulation model | none in the library | none in the library |
The parts JX CLS6336, T-Motor MN5212 KV340, ODrive 24V and AS5047D have catalog rows and BOM prices, and no fact rows. Until those are collected, treat any torque claim about these robots as unchecked here.
Other open quadruped entries do not fill this gap either. ODRI Solo has URDF rows and no BOM rows. Pupper v3 is a separate catalog entry with a URDF row and no BOM rows. ODRI Actuator Module v1 has 34 unpriced rows around a T-Motor Antigravity 4004. Its fact rows claim 9:1 total reduction and 2.5 N·m at 12 A, but they describe that module, not Pupper or Doggo.
Safety rows
Doggo lists one relay under power distribution: "Solidstate Relay (Replaces minitactor)", $137.92. Pupper lists no switch, relay or fuse row. Neither list has a fuse or breaker row, so the fuse-or-breaker check is open for both [8]. These lists may not be complete; the database only shows what the rows say.
Guidelines this example exercises
| Guideline | Result here |
|---|---|
| A servo's stall torque is its maximum momentary output and differs from its continuous rating. Real-world output is generally closer to the performance-graph measurements [7]; In continuous operation keep motor torque below the thermally set continuous limit, not the peak or stall value. [9] | Cannot run: no torque values. |
| Quasi-direct drive uses a single-stage reduction with a ratio under 10:1. It raises torque at the expense of control bandwidth but keeps the motor backdrivable [1]; Low gear ratios reduce reflected inertia but can increase Joule-heating energy for a given torque [4]; With motor mass scaling with radius, torque scales with radius squared and reflected inertia with radius cubed [2]; Expect outrunner motors to trade reduced thermal dissipation and higher inertia for higher torque density. [3] | Explain the Doggo choice. Doggo's ratio is not in the database, so the under 10:1 test is not run. |
| Confirm the servo amplifier delivers both the average and the peak motor current [10]; A multi-cell battery must provide the required capacity and current within the application's voltage limits. [11] | Cannot run: no current values. |
| Doggo runs field-oriented commutation at 10 kHz; treat this as an example, not a limit [6]; Doggo's motor test raised current until the winding reached a steady 125 C, a value often used as the maximum for stator windings. [5] | Doggo paper values, as examples. |
| Protect every power wire with a fuse or breaker located as close as possible to the source bus. [8] | Open: no fuse or breaker row on either list. |
Sources
Each guideline in this chapter with its source, the date it was read and the exact words of the source. Level: must means breaking it causes failure or harm, should is good practice, guide is a rule of thumb.
- Quasi-direct drive uses a single-stage reduction with a ratio under 10:1. It raises torque at the expense of control bandwidth but keeps the motor backdrivable. Level: guide. arxiv.org 2026-11-08
- With motor mass scaling with radius, torque scales with radius squared and reflected inertia with radius cubed. Level: guide. arxiv.org 2026-11-08
- Expect outrunner motors to trade reduced thermal dissipation and higher inertia for higher torque density. Level: guide. arxiv.org 2026-11-08
- Low gear ratios reduce reflected inertia but can increase Joule-heating energy for a given torque. Level: should. fab.cba.mit.edu 2026-11-08
- Doggo's motor test raised current until the winding reached a steady 125 C, a value often used as the maximum for stator windings. 125 C Level: guide. arxiv.org 2026-11-08
- Doggo runs field-oriented commutation at 10 kHz; treat this as an example, not a limit. 10 kHz Level: guide. arxiv.org 2026-11-08
- A servo's stall torque is its maximum momentary output and differs from its continuous rating. Real-world output is generally closer to the performance-graph measurements. Level: guide. emanual.robotis.com 2026-11-08
- Protect every power wire with a fuse or breaker located as close as possible to the source bus. Level: must. raanz.org.nz 2026-11-08“with circuit breakers or fuses located as close”
- In continuous operation keep motor torque below the thermally set continuous limit, not the peak or stall value. Level: must. hades.mech.northwestern.edu 2026-11-08
- Confirm the servo amplifier delivers both the average and the peak motor current. Level: must. maxonmotor.com 2026-11-08
- A multi-cell battery must provide the required capacity and current within the application's voltage limits. Level: must. ntrs.nasa.gov 2026-11-08“adequate thermal management. The battery must provide both the capacity and current required”